Undergraduate Research Opportunities Program (UROP) Proceedings 2020-21

Abstract



Table of Contents Message from the President i Sharing by Students ii-iii UROP Overview 2020-21 iv Mr. Armin and Mrs. Lillian Kitchell Undergraduate Research Awardees 2021 v-vi Abstracts of UROP Projects - 2020-21* School of Science (CHEM, LIFS, MATH, OCES, PHYS) 2-77 School of Engineering (CBE, CIVL, CSE, ECE, IEDA, ISD, MAE) 80-183 School of Business and Management (ACCT, ECON, FINA, ISOM, 186-218 MARK, MGMT) School of Humanities and Social Science (HUMA, SOSC) 220-236 Interdisciplinary Programs Office (ENVR) 238-245 Institutes & Centers (IEMS) 248-249 *Abstracts from each school are listed first by alphabetical order of Department code, and then by alphabetical order of Advisor’s surname. Summary of UROP Courses UROP 1000 Undergraduate Research Opportunities Program (0 credit with stipend option, offered in summer sessions only) UROP 1100 Undergraduate Research Opportunities Program Series 1 (1 credit, offered throughout the year) UROP 2100 Undergraduate Research Opportunities Program Series 2 (1 credit, offered throughout the year; prerequisite is pass in UROP 1100, with approval by project advisors) UROP 3100 Undergraduate Research Opportunities Program Series 3 (1 credit, offered throughout the year; prerequisite is pass in UROP 2100, with approval by project advisors) UROP 4100 Undergraduate Research Opportunities Program Series 4 (1 credit, offered throughout the year; prerequisite is pass in UROP 3100, with approval by project advisors) Message from the President The year of 2021 marks a significant milestone for HKUST as we are celebrating our 30th anniversary. We open a new chapter in the history of HKUST where we will continue exploring and experimenting for societal impact. The Undergraduate Research Opportunities Program (UROP) has played an indispensable role in providing our undergraduate students hands-on experience in academic research, which has empowered creativity and innovation. This well represents our 30th anniversary theme of “Together We Empower”. As we embark on a new academic year under the continuing challenges of the COVID-19 pandemic, our dedicated faculty members remain steadfastly committed to providing our undergraduate students exceptional research and learning experiences. Their efforts help boost the research culture and nurture young researchers within the university community. During the 2020-21 academic year, 160 faculty members served as supervisors for 645 students. I am thankful for all the participating faculty members and undergraduate students who have devoted considerable time and efforts in their research projects and contributed to another record high in the UROP enrolment. I trust that all UROP students have had a rewarding research experience while pursuing their passions in diverse academic frontiers. These proceedings highlight the research projects accomplished by our undergraduates under the guidance of their faculty supervisors. I am hopeful that more of our students will join us and immerse themselves in the wonderful journey of discovery. Professor Wei SHYY President HKUST i Sharing by Students NG Cheuk Hei BSc in Biochemistry and Cell Biology

My experience in my supervisor’s lab was very fruitful under the UROP program. It enriched my understanding on protein trafficking and equipped me with relevant experimental knowledge. I joined research with the sole intention to learn and explore new fields, but during the process of exploring, what I got was way more than I had expected. In addition to the intriguing experience of learning on a given topic, I also had the opportunity to take more control by designing an experiment and exploring ideas of my own. The UROP program offered me much more than knowledge and skill - I was able to experience research firsthand and see how imagination is my only limit. NGUYEN Ha Chau BSc in Biotechnology

UROP has brought the field of research much closer to me. At the beginning, I wanted to join UROP to have a taste of research, but later I realized that research is what I want to pursue for my career. Thanks to numerous opportunities and flexible requirements, I can work in a lab on a research topic of my interest without worrying which department that lab belongs to. As a Life Science student working in a Chemical and Biological Engineering lab, I feel lucky since I can balance between gaining basic Life Science knowledge and grasping the practical aspect of it. UROP has helped me form a solid foundation to continue pursuing applied biology in the future. ii

REN Xuanchi BSc in Computer Science

Thanks to UROP, I have had a great chance to work with some helpful professors and begin my first research project on cutting-edge technology. It has been an excellent opportunity for me to learn how to select a research topic, how to review research papers, how to design and conduct experiments, and how to convert an idea into a paper. Furthermore, participating in research during undergraduate study has enabled me to understand the basics of academic research before moving onto postgraduate study. UROP has inspired me to make continuous effort in research, which is important for me to make my career decisions in the future. MAK Ting Hin BSc in Physics

UROP has offered me the opportunity to gain a clearer picture on what academic research is like, by having hands-on experience under supervision in various tasks that an academic researcher would come across with, such as performing experiments, analyzing experimental results, participating in group meetings, writing reports and papers, giving presentations to different target audiences and making posters. It is also a valuable experience for me to practice essential soft skills for becoming a researcher, such as critical thinking and collaboration skills. All these experiences are very helpful to me for preparing my future research career. iii UROP Overview 2020-21 In the 2020-21 academic year, the Undergraduate Research Opportunities Program (UROP) has seen a substantial growth in participation compared to the previous year. Over 230 faculty supervisors from four Schools and from the Interdisciplinary Programs Office offered around 600 projects for our undergraduates. We also received more than 1,000 applications throughout the academic year, with 645 students were successfully enrolled to the program. To enhance the continuous development of existing UROP projects, the UROP Support Grant Scheme continued providing financial support for faculty supervisors and their students. Through the Scheme, students also experienced the process of applying for research grants by submitting a joint application with their supervisors. In 2020-21, 41 out of 46 received applications were awarded with an aggregate funding amount of over HK$700,000, after the review by the UROP Office and by the UROP Advisory Board. The 2021 Mr. Armin and Mrs. Lillian Kitchell Undergraduate Research Award and the UROP Faculty Research Award In the year 2021, 25 students with outstanding research performance were nominated by their UROP supervisors for the Mr. Armin and Mrs. Lillian Kitchell Undergraduate Research Award. Among the nominees, 11 candidates were shortlisted and invited to give an oral presentation on their research findings to the UROP Advisory Board. The Board recommended 1 Champion, 2 First Runner-Ups and 3 Second Runner-Ups. The faculty supervisors of those 6 student awardees were also recognized also recognized by the UROP Faculty Research Award. The award presentation ceremony was held via Zoom on 23 April 2021. iv Mr. Armin and Mrs. Lillian Kitchell Undergraduate Research Award 2021 List of Awardees Champion MAK Ting Hin Major / Year: PHYS-IRE / 3 Supervised by: JO Gyu Boong / PHYS Project Title: Machine Learning Aided Detection of the Quantum State in an Atomic Quantum Simulator First Runner-up CHEN Yijia Major / Year: COSC / 3 Supervised by: CHAU Kevin / ECE Project Title: A Real-Time Wavelet-Based Algorithm for Improving Speech Intelligibility First Runner-up & Best Poster Award REN Xuanchi Major / Year: COSC / 4 Supervised by: CHEN Qifeng / CSE Project Title: Self-supervised Dance Video Synthesis Conditioned on Music Second Runner-up v KWOK Tsz Hong Major / Year: PHYS / 4 Supervised by: WONG Michael Kwok Yee / PHYS Project Title: Neural Information Processing: Latency Prediction in the Highway System Second Runner-up NG Cheuk Hei Major / Year: BCB-IRE / 3 Supervised by: GUO Yusong / LIFS Project Title: Analysis of SARS-CoV2 Spike Protein and ACE2 Interaction Second Runner-up NGUYEN Ha Chau Major / Year: BIOT / 3 Supervised by: SUN Fei / CBE Project Title: Spectral Tuning of Water-Soluble Chlorophyll Binding Proteins for Deep Tissue Bioimaging vi Abstracts of UROP Projects 2020-21 School of Science School of Science Department of Chemistry Department of Chemistry Construction and Application of Surface Enhanced Raman Spectrometer in Biomolecules Characterization Supervisor: HUANG Jinqing / CHEM Student: CHAU Cheuk Yui Sherman / BICH Course: UROP1000, Summer With its highly distinct, non-destructive characterization of chemical samples, Raman spectroscopy is utilized in a plethora of fields, including pharmaceuticals, cosmetics, and geology. Its rapid identification of samples could therefore prove useful in differentiating between cancerous and non-cancerous breast tissue from the same patients. This report presents the application and viability of surface enhanced Raman spectroscopy in the characterization of biomolecules. The results provide evidence that the biochemical composition of tumor, papilloma, and fibroadenoma notably different but contain a similar spectra pattern. To analyse the spatial configuration of the tissue sample, Raman mapping is also utilized to visually identify the discrepancy between cancerous and noncancerous breast tissue and its influence on Raman scattering. Construction and Application of Surface Enhanced Raman Spectrometer in Biomolecules Characterization Supervisor: HUANG Jinqing / CHEM Student: SUN Weixuan / CHEM Course: UROP1100, Summer Cancer poses a serious threat to modern people health, so rapid detection of cancer is very important. This project focus on using multiple surface enhanced Raman spectrometers to scan different cancer tissue samples then obtain a series of Raman spectra. The purpose is to detect the special characteristic peaks of cancer tissues from the spectra. Analyzing soft wares, such as Origin Lab, are used to rebuild the curve from excel files and process data to build new curves. Then we use a mapping tool as the result has no distinctive features. A normal sample without cancer cells is used as a control group. Single Molecule Manipulation and Characterization of Biomolecules Supervisor: HUANG Jinqing / CHEM Student: SALIM Leon Ritchie / CHEM Course: UROP1000, Summer Scientists have used bulk techniques, such as Raman Spectroscopy and Infrared Spectroscopy, to analyse chemical compounds. However, there are limitations on only using these techniques. There is a need to analyse molecules thoroughly by itself, and this is where Single Molecule Force Spectroscopy comes in. It allows us to meticulously study not only the unfolding, but also the transition states of the biomolecules and the forces in the transitions. This report introduces the Single Molecule Force Spectroscopy and their examples. It also includes the background of optical tweezers and DNA properties. Finally, this report ends with short and long DNA optical tweezers results, the fittings using MATLAB and their analysis. 2 School of Science Department of Chemistry Single Molecule Manipulation and Characterization of Biomolecules Supervisor: HUANG Jinqing / CHEM Student: LEE Dong Heon / CHEM Course: UROP1100, Summer The single-molecule study opens a new window to investigate molecular structures and their chemical properties. In this project, we focused on DNA stretching as it is widely used as a handle for optical tweezer experiments. The short and long DNAs are extended via optical tweezer then, their concordance rate and fitting parameters are derived. The results interpretation confirmed the current drawbacks of single-molecule experiments. The short and long DNA experiments exhibited that the sample molecule can interfere with measurements, and utilized extensible WLC may be deficient in modeling the phenomenon. Further, by comparing the data of short and long DNA, we found that the stretching geometry or length of the molecule can also contribute to the difference in the result. Medicinal Chemistry on Novel Type I1/2 ALK Inhibitors for Combating Drug-resistant Mutants Supervisor: HUANG Yong / CHEM Student: MA Wing Yin Timothy / CHEM Course: UROP1100, Fall Anaplastic Lymphoma Kinase is a tyrosine kinase which is expressed in different types of cancer tumor. In non-small cell lung cancer (NSCLC), ALK gene loci on chromosome 2 has caused rearrangement in about 5% of NSCLC tumors. Detecting ALK gene and using suitable therapy is critical in treatment of lung cancer. Anaplastic Lymphoma Kinase (ALK) inhibition is a therapy targeting the ALK-positive tumors and preventing the expressing of this oncogene. This project is aimed to synthesis an ALK inhibitor molecule. Medicinal Chemistry on Novel Type I1/2 ALK Inhibitors for Combating Drug-resistant Mutants Supervisor: HUANG Yong / CHEM Student: WONG Tsz Wang / CHEM Course: UROP2100, Fall It is proposed that the meta, para selective borylation of aromatic esters and amide can be controlled using Ir-catalysis with different ligand in various length, shape and directing group. In the previous project, we have design 4 ligands and 2 of them have been synthesized. In this project, the remaining ligands are synthesized using optimized condition and the Selective CH Remote Borylation of Benzoxazole is studied using quaterpyridine ligand, L4. Nuclear Magnetic Resonance (NMR) Spectroscopy, Gas Chromatography/Mass Spectrometer (GC/MS) are using to confirm the quality, quantity, and purity of the synthesized compounds. 3 School of Science Department of Chemistry Chemistry of Aromatic Metallacycles Supervisor: JIA Guocheng / CHEM Student: HO Hang Chi / SSCI Course: UROP1100, Summer Osmabenzynes are a family of organometallic complexes containing a benzyne ring with an osmium atom. Organic benzyne is a highly reactive species due to ring strain, which is why it has never been isolated. However, with the substitution of one of the carbons with an Os atom, the complex becomes much more stable. The osmaanthracyne complex (an organometallic analogue of anthracene), the synthesis and reactivity of which will be the focus of this report, is air stable and has been well-characterised. The remarkable stability of osmaanthracyne complex can be attributed to the very large size of the Os atom compared to C, and the steric hindrance caused by the large size of the trimethylsilyl group (TMS), thus protecting the electron-rich benzyne from electrophilic attack. The objective of this study is to study the similarities between anthracene and its organometallic analogue: osmaanthracyne complex. Chemistry of Aromatic Metallacycles Supervisor: JIA Guocheng / CHEM Student: YEUNG Tsz Ho / SSCI Course: UROP1100, Summer Metallabenzynes have much to be studied and discovered. The Jia group was the first to synthesize a metallabenzyne complex. With unsubstituted osmanaphthalynes and osmaanthracynes successfully synthesized by the group before, an unsubstituted osmabenzyne has yet to be synthesized. Taking part in synthesizing and characterizing an unsubstituted osmabenzyne in my UROP1100 project, I successfully synthesized the osmabenzyne 8 with a -TMS group independently. During my experiments, I am fortunate enough to take part in uncovering new discoveries about the synthetic route of the osmabenzyne 8. The following report documents my experimental results and data from my independent synthesis of osmabenzyne 8, as well as discussion on such results. Mentions on future works and the experimental information will also be included. 4 School of Science Department of Chemistry Synthesis, Reactivity and Catalytic Properties of Transition Metal Carbyne Complexes Supervisor: JIA Guocheng / CHEM Student: HO Tung Ho / SSCI Course: UROP1100, Summer The non-d0 carbyne complexes are difficult to undergo alkyne metathesis. We aim to find an active rhenium carbyne catalyst for alkyne metathesis and test for their activities. We are using an organometallic compound (Rhenium) to have alkyne metathesis with different catalysts. This work demonstrates the complex Re(≡CH2Ph) (PO2) (PMePh2) 2 to react with PhC≡CMe to produce Re(≡CPh) (PO2) (PMePh2) 2 and also Re(≡CH2Ph) (SO2) (PMePh2) 2 to react with PhC≡CMe to produce Re(≡CPh) (SO2) (PMePh2) 2 to check whether they are suitable rhenium carbyne catalysts. It will also show how we synthesis different reactants needed for the metathesis. We will check the correct structure by using Nuclear Magnetic Resonance (NMR) using phosphorus-31 due to the large molecule containing phosphorus. Tetravalent Cerium Complexes Containing Oxygen Donor Ligands Supervisor: LEUNG Wa Hung / CHEM Student: WANG Lily Ueh-hsi / CHEM Course: UROP1100, Spring A tetravalent cerium sulfamate complex supported by the Kläui tripodal ligand [CpCo{P(O)(OEt)2}3]- (LOEt - ) has been synthesized and its redox reactivity toward 2,6-di-tertbutylphenol was studied. The reaction of [CeIV(LOEt)2Cl2] and AgSO3NH2 afforded [CeIV(LOEt)2(OSO2NH2) 2], whose crystal structure was determined. Plausible synthetic routes to tetravalent cerium selenite and tellurite complexes are still under investigation. Reactions starting from Ag2CO3 seemed to be an issue since the carbonate residue in salt preparation led to the formation of [CeIV(LOEt)2(OCO2)] which is a previously reported complex. The reactions between [CeIV(LOEt)2(OAc)2] (OAc- = acetate) and [CeIV(LOEt)2X2] (X = Cl- , NO3 - , OTs- ) were investigated by NMR spectroscopy and preliminarily results showed that ligand exchange between these complexes occurred. Application of Molecular Orbital Theory to Transition-Metal Complexes Supervisor: LIN Zhenyang / CHEM Student: WONG Wing Hei Marco / CHEM Course: UROP1100, Summer With reference to the work of Yang, Sheong and Lin, this work studied coordination of different ligands to (MesCCC)CoN2 (MesCCC=bis(mesityl-benzimidazol-2-ylidene)phenyl) and (PCP)CoN2. The relative cis-trans stabilities of (MesCCC)Co(H)2 and (PCP)Co(H)2 were also systematically evaluated. Coordination of ligands having strong π accepting and weak σ donating abilities at the apical site of square pyramidal Co(I) pincer complexes was found to be favourable. Phosphine coordination to (MesCCC)CoN2 is more favourable than that to (PCP)CoN2. The empty 2p orbitals of carbenes and empty 4p(Co) orbital of (MesCCC)CoN2 together stabilize phosphine coordination. The delocalized MO formed among the aforementioned orbitals accounts for the higher stability of trans-(MesCCC)Co(H)2 than its cis-isomer, too. In contrast, cis-(PCP)Co(H)2 is more stable than its trans-isomer. It is anticipated that this work may enhance our understanding on the chemistry of carbene pincer ligands. 5 School of Science Department of Chemistry Deep Learning in Synthesis Planning Supervisor: SU Haibin / CHEM Student: CHIU King Wai / CHEM Course: UROP1100, Spring Machine learning in Chemistry becomes hot topic currently. Computer-aided synthesis planning (CASP) with retrosynthesis can be constructed for chemists to search for plausible most effective and cheapest synthesis route with machine learning technique, such as Artificial Neural Networks (ANN) and Monte Carlo Tree Search (MCTS). However, commercial database like Reaxys cannot provide accurate and detailed chemicals and reaction condition, especially when organometallic reaction is considered. In this project, automatic data-mining tools are constructed with Optical Structure Recognition Software (OSRA) and Optical Character Recognition (OCR) Software which utilized Long Short-Term Memory (LSTM) model. In future, there is high potential to build a machine learning model in chemistry paper reading to further improve the data-mining. Important reaction elements, such as catalysts, ligands, additives, temperature and yield can be extracted to enrich the database to outperform the current available commercial and open source and be treated as more detailed descriptors to feature the reaction rules for better retrosynthesis planning by SPLASH. Deep Learning in Synthesis Planning Supervisor: SU Haibin / CHEM Student: CHUI Sin Yu / CHEM Course: UROP1100, Spring Conditions of reactions are important to determine and proceed synthesis planning that is comprehensively applied in many fields of chemistry. Reaction data analysis may help chemists to make decisions whether the pathway would be high-efficient without on-hands laboratory work. Utilizing which type of catalyst is also one critical condition required in catalytic reactions for organic synthesis. This report is an overview of reactions using nickel catalyst with bidentate nitrogen ligands by reaction data analyzing. Nine types of reactions are covered including Suzuki coupling, Negishi coupling, Hiyama reaction and C-H activation with the usage of 21 ligands like BBBPY, bpy, BPhen and (R,R)-ph-box. Deep Learning in Synthesis Planning Supervisor: SU Haibin / CHEM Student: FUNG Ka Shing / DSCT Course: UROP1000, Summer Retrosynthesis is one of the hottest topics in chemistry. Our aim is to build a retrosynthesis program with the help of machine learning. SMILES and SMARTS, the notation of compounds briefly explained in computer. Monte Carlo Tree Search and artificial neural network, computer software RDChiral, Reaction Decoder are applied. A neural network was trained and verified by testing. Two compounds, curcumin and vortioxetine were put into the model, successfully generated some positive results. There were a few failure cases due to reaction centre selectivity and ring closing problem. Further improvements could be made by importing more conditions into the neural network and tree searching criteria. 6 School of Science Department of Chemistry Deep Learning in Synthesis Planning Supervisor: SU Haibin / CHEM Student: CHUI Sin Yu / CHEM LEE Subin / CHEM WANG Yizhou / CHEM Course: UROP2100, Summer UROP1100, Summer UROP1100, Summer Conditions of reactions are crucial to determine and proceed synthesis planning that is comprehensively applied in many fields of chemistry. Reaction data analysis may help chemists to decide whether the pathway would be more high-efficient without on-hand laboratory work. Utilizing which type of catalyst is also one important condition required in catalytic reactions for organic synthesis. This report is an overview of reactions using copper catalyst with bidentate nitrogen ligands by reaction data analyzing. General classes such as C-C bond formation, reaction types and the usage of bidentate nitrogen ligands are included. Employing Coevolution Analysis for Investigating and Improving Cas9 Specificity Towards More Safe CRISPR-Based Genome Editing Supervisor: SU Haibin / CHEM Student: WOO Dorothy Hoi Shan / BTGBM Course: UROP1100, Fall UROP2100, Spring CRISPR is a natural adaptive immune system found many naturally occurring organisms. While there are many proteins acting as an effector module in CRISPR, CRISPR-associated protein 9, more commonly known as Cas9, has shown great efficiency to be used for targeted genome editing. The CRISPR/ Cas9 system has been applied to many biological research areas, such as gene therapy, bioimaging and drug discovery. However, off-targeting behaviour and specificity of these proteins have not been fully understood, and a more extensive understanding towards the specificity of Cas9 mechanisms is needed to improve genetic engineering and editing techniques. This progress report devotes itself to utilizing the approach of coevolution analysis to study the relative interactions of proteins in the Cas9_BH and HNH_4 domains. Through coevolution analysis, we identified the amino acid residue couplings with high correlation in the two domains respectively, and data can be utilized in further research to study the specificity of Cas9 activity in these domains. 7 School of Science Department of Chemistry Employing Coevolution Analysis for Investigating and Improving Cas9 Specificity Towards More Safe CRISPR-Based Genome Editing Supervisor: SU Haibin / CHEM Student: TANDIAWAN Cathleen Trixie / SSCI Course: UROP1000, Summer The worldwide spread of SARS-CoV-2 has drawn attention around the globe as it significantly influences humans, to an extent that it changes the way of living. Recombinant and vaccine-induced neutralizing antibodies have been made to tackle the spread of this virus. However, the newly emerged lineage, B.1.351, much known as beta variant from South Africa is raising concerns as it is speculated to have higher resistance. We evaluated each mutation of concern from this variant to see how the lineage surged up. Through elaborate mutation tracking, the buildup can be explained including mutations that might increase its severity. Furthermore, to the best of our knowledge, throw light on new takes to develop a cure, still in efforts to restrain the pandemic. Employing Coevolution Analysis for Investigating and Improving Cas9 Specificity Towards More Safe CRISPR-Based Genome Editing Supervisor: SU Haibin / CHEM Student: YAP Shan Qi / CHEM Course: UROP1000, Summer Mutations in the spike protein of SARS-CoV-2 have allowed successful variants to emerge and rise to prominence. However, the reason behind these mutations’ emergence and the ways in which they evolve still remain unclear. By using spike protein sequences of SARS-CoV-2 downloaded from GISAID, we intend to look at the trajectories of mutations in the spike. We obtained a better understanding of the mutations by tracing each mutation established in the P.1 (Gamma) variant. It is suggested that the mutations are linked to viral fitness, leading up to the variant’s survival and persistence. We hope that our findings could be useful in helping to predict important mutations that might arise in the future. Prediction of Inorganic Solar Cell Materials with Double Perovskite Structure Using Machine Learning Approach Supervisor: SU Haibin / CHEM Student: CHEUNG Ka Key / CHEM KWAN Yee Tung / SENG Course: UROP1000, Summer UROP1000, Summer Partial charge determination has been playing an important role in chemistry. A new method of partial charge determination, the modified Born-Haber cycle, has proposed recently. In order to verify the method, M(I)X, which is composed by group 1 and group 17, has been calculated and compared with experimental data in the published paper. However, it is insufficient enough to make it a reliable method. Hence, our project is designed to further prove the method through applying it to the group 2 & 16 (M(II)X), group 13 & 15 (M(III)X) and group 14 & 16 (M(IV)X) diatomic system. Since most of the M(III)X and M(IV)X molecules exist theoretically, Gaussian 09 B3LYP/QZVP has been used for modelling. 8 School of Science Department of Chemistry Structure Prediction with Genetic Algorithm Supervisor: SU Haibin / CHEM Student: ZOU Zhiyan / CHEM Course: UROP1100, Fall Synthesis planning is one of the research approaches at the frontier of chemistry, as well as medicine and pharmacy. Our research focus on the machine learning models in retrosynthesis systems, making it necessary the extraction scaffold reaction information. The process follows a well design workflow. After scaffold information extraction, reactions are clustered by their forming bonds and ligand classes in order to find the relationship between reactions and reaction types. At current progress of this project, we only focus on the single-step Nickel-catalysed reactions. Valuable properties and hidden trends for Cross-Coupling and addition reactions emerged for further analysis and ready for deep learning in synthesis planning. Development of New Catalytic Organic Processes Supervisor: SUN Jianwei / CHEM Student: YAN Qiaolin / CHEM Course: UROP1100, Spring The work of two subprojects is discussed in this report. In the first project, various catalysts were screened in the hope to selectively oxidize 1,2,4-trimethylbenzene to 2,3,5-trimethyl-1,4-benquinone or 2,3,5-trimethylhydroquinone using hydrogen peroxide as a mild and green oxidant. Synthesis of vitamin E from 2,3,5-trimethylhydroquinone was also attempted. In the second project, different hydride sources and various chiral phosphoric acids were screened for the enantioselective synthesis of triarylmethane via the reductive dehydroxylation of triarylmethanol. Preparation of the identified optimal catalyst and further optimization of the reaction conditions is still ongoing. Development of New Catalytic Organic Processes Supervisor: SUN Jianwei / CHEM Student: LI Yuxuan / SSCI Course: UROP1000, Summer The direct enantioselective 1,6-addition of 5-methide-5H-indoles, generated in situ from corresponding diarylmethanols, has been attempted in the presence of chiral phosphoric acids (CPAs). Preliminary experiments using various nucleophiles with racemic phosphoric acid (RPA) have been carried out to select feasible reaction partners of diarylmethanols. To achieve the remote activation of the substrate and the enantioselectivity of the substitution reaction, evaluation of catalysts, solvents, concentration of reactants, and temperature have been performed. Furthermore, the modification of the diarylmethanols’ structure has also been attempted in hopes of achieving better enantioselectivity. Although such a reaction mode is currently not synthetically applicable because the enantioselectivity (ee value up to 56%) requires further enhancement, the efforts in condition optimization and substrate modification could be valuable for investigation of similar reactions. 9 School of Science Department of Chemistry Development of New Catalytic Organic Processes Supervisor: SUN Jianwei / CHEM Student: YAN Qiaolin / CHEM Course: UROP1000, Summer Enantioenriched triarylmethanes with heterocyclic structures are critical components in drugs. Described here is a new approach to access this important and challenging target in organic synthesis via chiral phosphoric acid-catalyzed reductive dehydroxylation of tertiary alcohols. This study presents the first reported success of a directing-group-free design in the asymmetric catalysis of triarylmethanes. The suitable chiral phosphoric acid catalyzed the mild reductive dehydroxylation to form diverse heterocycle-containing triarylmethanes with excellent efficiency and enantioselectivity. Control experiments, kinetic study, and non-linear effects revealed the kinetic resolution in the rate-determining carbocation formation step to determine the enantioselectivity. Design and Synthesis of Functional AIE luminogens and Exploration of their Biological Applications Supervisor: TANG Benzhong / CHEM Student: WANG Lily Ueh-hsi / CHEM Course: UROP1100, Fall With the discovery of aggregation-induced emission, materials leveraging some distinct features in the aggregate state have been flourishing in the past twenty years. Cocrystal, with two or more compounds aggregate by noncovalent interactions, is an arising field of great versatility in terms of tunable physicochemical properties. To develop materials with photothermal properties, self-assemble cocrystal made from simple preparation is one promising direction. In 2018, the first cocrystal with the ability of near-infrared and heat conversion came to place. Here, inspired by the recent breakthrough in aggregate science, a cocrystal is reported, which successfully demonstrated photothermal properties in the preliminary investigation. Design and Synthesis of Functional AIE luminogens and Exploration of their Biological Applications Supervisor: TANG Benzhong / CHEM Student: LI Sijie / CHEM Course: UROP1000, Summer Aggregation-induced emission (AIE) luminogens (AIEgens) are a type of molecule showing emission under aggregate state. The powerful emission in aqueous media, a common poor solvent for organic compounds, demonstrates a wide application of this novel type of molecule. AIE-active moiety TBP has been reported for the ability to generate reactive oxygen species (ROS) for antibacterial and antitumor effect. Whereas quinone methide (QM) has been found to be effective in inhibiting antioxidant system for resisting ROS-induced apoptosis in cancer cells. Therefore, in this study, a synergistic hybrid AIEgen TBP-QM is designed and synthesized for cancer treatment. TBP-QM is expected to act as a novel theranostic agent with the ability to disturbing cancer cell antioxidant system, inducing cell death by ROS invasion in a site-specific way, and at the same time allowing on-site bioimaging. 10 School of Science Department of Chemistry Design and Synthesis of Functional AIE luminogens and Exploration of their Biological Applications Supervisor: TANG Benzhong / CHEM Student: ZHAO Mengying / CHEM Course: UROP1000, Summer Nonconventional small luminophores lacking remarkable conjugations may exhibit excitation-dependent photoluminescence (PL) properties, which are generally considered to stem from the diverse aggregation or clustering of the non-conjugated electron-rich moieties. The phenomenon is termed clusteroluminescence, which can be rationalized through a clustering-triggered emission (CTE) mechanism. CTE has proven its great potential in designing and constructing novel, easily tunable luminophores in a variety of fields, including bioimaging, chemical sensors, and biomolecule conformation indicators. Based on the CTE mechanism, in this work, three originally nonemissive, nonconjugated molecules were reevaluated and clusteroluminescence was observed in the aggregated and solid states of the three molecules respectively. Upon the excitation of near-ultraviolet lights in different wavelengths, the emission wavelengths of the three molecules were varying from 387nm to 450nm. Extensive intermolecular H-bonding was proposed to elucidate the unusual clusteroluminescence. These findings will decode the working principle of nonconventional luminophores and facilitate a deeper understanding of color-tunable clusteroluminogens. Chemical Characterization of Atmospheric Pollutants Supervisor: YU Jianzhen / CHEM Student: LIU Sum Yee Roxanne / CHEM Course: UROP1100, Fall In this paper, high-performance anion-exchange chromatograph (HPAEC) with detection by pulsed amperometric detection (PAD) is used to detect and analyse monosaccharide anhydrides (MA) that are present in air such as levoglucosan and mannosan. The aim of the project is to do online analysis of MA with a low limit of detection so as to detect the scarce concentration of MA. However, the length of the report period is not adequate to accomplish the aim. During the report period, we mainly do offline analysis of the standard solutions of MA as well as develop the instrumentation for online analysis of MA. 11 School of Science Division of Life Science 12 Division of Life Science Studying Novel Cell Cycle Signalling Pathways in Yeast Cells Supervisor: BANFIELD David Karl / LIFS Student: SINCE Alec Victor / BIBU Course: UROP1100, Summer The purpose of this study is to investigate the stress-induced ectopic activation of the Spindle Assembly Checkpoint (SAC). We investigate the kinases of the cell wall integrity (CWI) pathway and determine whether they are involved in SAC activation. One of these kinases is Pkh1p and the deletion of this kinase seems to delay SAC activation. We came to this conclusion by using Mad2p-mNeon fluorescence microscopy imaging and calculated the proportion of cells that form Mad2p kinetochore foci. Further studies are required to determine whether the deletion of Pkh1p has a direct or indirect effect on delaying SAC activation or whether the Pkh2p homolog of Pkh1p can replace its function in its absence. Molecular Regulation of Skeletal Muscle Stem Cell Quiescence and Activation Supervisor: CHEUNG Tom / LIFS Student: LI Chun Wa / BCB Course: UROP1100, Fall Muscle satellite cells (MuSCs) perform a crucial role in skeletal muscle regeneration after injury. In this UROP 1100, emphasis is put on analysis on differential expression of these cells in post injury. Using the Galaxy platform, RNA-seq was performed using the Tuxedo pathway and Gene Set Enrichment Analysis (GSEA) using EGEA, which identified 7174 differentially expressed genes (DEGs) in 20367 loci. Via EGSEA, it was detected activation in molecular pathways for proliferation and differentiation such as E2F-dependent cell cycle transition, G2M checkpoint, and mTORC1 signalling. These data suggested that post-injured satellite cells may exit quiescent state and carry out proliferation after injury, possibly for regeneration of damaged skeletal muscle cells. Construction of a Signal Transduction Pathway Reporter Indicator for Monitoring Signaling Strength Supervisor: CHOW King Lau / LIFS Student: YUEN Yan Yi Macy / BIOT Course: UROP1100, Spring This project will manipulate the genetic model system, Caenorhabditis elegans, to recruit one of the novel gene components regulated by the transforming growth factor-β, DBL-1, as an indirect indicator of the BMP signaling strength. Sma-6, the target gene for the project, is the type I receptor for DBL-1 ligands in the BMP pathway and is transcriptionally regulated by the dbl-1 signaling. Thus, the gene expression of sma-6 can reflect the intensity of the BMP pathway. This experiment will develop the sma-6 transcription reporter responding to the dbl-1 signaling and amplify the readout of signal strength with the use of several reporters. School of Science Division of Life Science Construction of a Signal Transduction Pathway Reporter Indicator for Monitoring Signaling Strength Supervisor: CHOW King Lau / LIFS Student: YUEN Yan Yi Macy / BIOT Course: UROP2100, Summer This project will manipulate the genetic model system, Caenorhabditis elegans, to recruit two of the gene components regulated by the transforming growth factor-β related ligand, DBL-1, as the indirect indicators of the dbl-1 signaling strength. Sma-6, one of the target genes for the project, encodes the type I receptor for DBL-1 ligands and is positively regulated by the dbl-1 signaling. The second gene, gcy-28, is a receptor-type guanylate cyclase and acts as a repression target to the BMP pathway. Thus, the gene expression of sma-6 and gcy-28 can reflect the intensity of the BMP transduction signal. This experiment will develop the sma-6 and gcy-28 transcription reporters responding to the dbl-1 signaling and amplify the readout of signal strength through the manipulation of these two reporters in a variety of dbl-1 mutant strains. Evolutionary Art at the Fast Track Supervisor: CHOW King Lau / LIFS Student: SIU Kwong Tai / PHYS Course: UROP1100, Fall In this study, we continue to learn the loss of antibiotics resistance of E.coli in various concentrations of antibiotics. By using chloramphenicol as our antibiotics sample, we have already shown that the loss of antibiotics resistant gene become significant when the concentration of antibiotics is greater than 50μg/ml for a 1-day growth. Base on this observation, we first perform similar measurements with a shorter time step. However, we reach an inconsistent observation. Thus we turn to work out the possible reason for causing this inconsistent and check whether we could reproduce the previous results when we wipe out the relevant factors. Evolutionary Art at the Fast Track Supervisor: CHOW King Lau / LIFS Student: LI Wa Hei / BTGBM Course: UROP1000, Summer With the extensive use of antibiotics and improper sewage treatment, antibiotic-resistant bacteria are a crucial problem worldwide even in the developed countries. Detection of mild concentration of antibiotics in rivers triggers the study of the effect of low antibiotic concentration on antibiotic-resistant plasmid retention in bacteria. In this study, we monitor the loss rate of the PBS-F21A3.3p- (4969 bp) plasmid in E. coli RR1 and it is concluded that the bacteria have highly preserved its antibiotic-resistant plasmid under 40 ng/ml of AMP. However, the effect of lower AMP concentration on plasmid retention requires further studies. 13 School of Science Division of Life Science The Role of Parkinson’s Disease Linked Gene Product Alpha-Synuclein in Mitochondrial Dysfunction Supervisor: CHUNG Kenny K / LIFS Student: CHO King Yi / BCB Course: UROP1100, Spring Parkinson’s disease (PD) is a severe yet common neurodegenerative disorder which is characterized by the gradual death of dopaminergic neurons in substantia nigra. Recent evidence has highlighted the relationship between the accumulation of alpha synuclein (aSyn) and mitophagy. Here, different level of WT aSyn was expressed in the neuroblastoma SH-SY5Y cell line. Carbonyl cyanide m‐chlorophenylhydrazone (CCCP) was used to induce mitophagy. For treated group (CCCP), there is an increased conversion of LC3-I to LC3-II along with increased level of aSyn. Besides, the increased degradation of mitochondrial protein COX 4 (cytochrome c oxidase IV) was also observed after CCCP treatment, suggesting potential cleavage after translocation to mitochondria. Therefore, the plasmids pRK5-COX4-pEGFP N1 and pRK5-Myc-COX4 were cloned for determining cleavage terminal. Analysis of Surface Delivery of Epidermal Growth Factor Receptors Supervisor: GUO Yusong / LIFS Student: NG Cheuk Hei / BCB Course: UROP3100, Fall The rapid spread of a novel and highly pathogenic coronavirus (SARS-CoV2) has caused a serious global public health emergency in 2020. The densely glycosylated spike (S) protein as a distinctive feature for corona viruses is the key player that promotes viral entry into cells. This protein is a trimeric protein composed of the S1 and S2 subunits. The receptor-binding domain (RBD), a subdomain of S1 directly binds a cell surface receptor, angiotensin-converting enzyme 2 (ACE2) present in human mucosal cells. This critical interaction for viral attachment serves as the prime therapeutic target. To develop effective therapeutic strategies, the interaction between ACE2 and CoV2-RBD, the essential step for viral attachment was analyzed to single residue resolution. With a pull-down assay, we have shown N501T and F486L to be key binding residues on CoV2-RBD in this interaction. Analysis of Surface Delivery of Epidermal Growth Factor Receptors Supervisor: GUO Yusong / LIFS Student: ZHOU Siyu / BCB Course: UROP3100, Spring Epidermal growth factor receptor (EGFR) is a tyrosine kinase which is activated by the binding of its specific ligands. Once activated, EGFR transfers from an inactive monomeric form to an active dimeric conformation which auto-phosphorylates its tyrosine residues in C-terminal domain and initiates several signal transduction cascades, leading to DNA synthesis and cell proliferation. Mutations that lead to EGFR overexpression have been associated with multiple cancer progressions, including head and neck, ovarian, cervical, bladder and esophageal cancers, as well as non-small cell lung cancer. These mutations lead to EFGR constant activation, causing uncontrolled cell division (Nicholson et al., 2001). My project aims to study whether the wild type and the cancer-related mutant versions of EGFR utilize the same machinery in the secretory pathway. 14 School of Science Division of Life Science Analysis of Surface Delivery of Frizzled6 Supervisor: GUO Yusong / LIFS Student: DANI Raissa Gavrila / BIOT Course: UROP1000, Summer This report investigated the function of SCAMP2 and Vti1b in vesicular trafficking from the Golgi Apparatus to the plasma membrane. SCAMP2 is a tetraspanning integral protein and Vti1b is a form of v-SNARE protein, both found in mammalian cells. We previously demonstrated siRNA knockdown procedures on both SCAMP2 and Vti1b separately. Post siRNA transfection, defects were observed for both SCAMP2 and Vti1b knockdown cells. Here, to further investigate whether or not the defects were the direct consequence of Vti1b knockdown and SCAMP2 knockdown, we generate rescue constructs for both genes that are resistant to siRNA knockdown. Both rescue constructs are generated using PCR mutagenesis by making primers with three base pairs mismatched on the siRNA annealing region. Since the experiment is currently in progress, we expected two possible outcomes after generating cell lines with the two rescue constructs. First, the defects will still be present indicating that the defects do not have a direct relationship to Vti1b or SCAMP2. On the other hand, if the defects disappear in the rescued Vti1b and SCAMP2 cell lines, this suggests that the defects indeed have a direct relationship to Vti1b and SCAMP2. Analysis of Surface Delivery of Epidermal Growth Factor Receptors Supervisor: GUO Yusong / LIFS Student: LI Chun Wa / BCB Course: UROP1000, Summer Epidermal growth factor receptor (EGFR) is a type 1 receptor tyrosine kinase involved in cell proliferation, differentiation, migration, and survival. As a vital receptor, the surface trafficking of this transmembrane protein is of great physiological significance. Rab12 is a potential mediator of EGFR transport with little attention. To study protein-proteininteraction of Rab12 with EGFR, immunoprecipitation can be applied for Rab12 purification, in which an affinity tag is required for RAb12 protein enrichment. Therefore, molecular cloning was done to generate FLAG tag at 3’ end of Rab12 gene based on Rab12-3xHA plasmid and pCMV-14 (3xFLAG) plasmid, followed by immunofluorescence staining to verify the normal functioning of Rab12- 3xFLAG protein by comparing with functionally-verified Rab12-3xHA. In this project, we successfully cloned Rab12-3xFLAG gene and undergone immunofluorescence imaging for verification. Analysis of Secretion of Insulin-like Growth Factor 2 Supervisor: GUO Yusong / LIFS Student: LEE Kai Ho / BICH Course: UROP1100, Summer TMED10 is one of the members of the P24 family proteins. In a previous study, it was found that TMED10 is important for Insulin-like growth factor 2 (IGF2) secretion, and hence may contributes to regulating skeletal myogenesis and myoblast differentiation. Meanwhile, TMED10 is composed of different domains, and the GOLD domain is suspected to play a role in specific cargo recognition. In this research, we want to study the importance of GOLD domain on IGF2 secretion by generating a TMED10 deletion mutant lacking the GOLD domain. Immunoprecipitation and RUSH-assay will then be conducted to see the effect of GOLD domain on the IGF2-TMED10 interaction and the secretion of IGF2. 15 School of Science Division of Life Science Molecular Mechanisms Regulating Secretion of Sonic Hedgehog Supervisor: GUO Yusong / LIFS Student: AGRAWAL Diya / BCB Course: UROP1100, Fall This report explored the trafficking of Frizzled6 protein using the RUSH system. Frizzled6 (Fzd6) being a transmembrane protein localised in the plasma membrane follow the secretory transport pathway. RUSH (Retention Using Selective Hooks) provides a mechanism to regulate the trafficking by retaining proteins of interest in desired organelles and releasing them upon treatment. In this report, RUSH-Fzd6 DNA construct is prepared by molecular cloning and transfected into HeLa cells. Fzd6 is tethered to Enhanced Green Fluorescence Protein (EGFP) as a reporter. Since the experiment is currently in progress, the paper establishes the expected outcomes of visualisation pattern of the protein under fluorescence microscope where initially, reporter and protein of interest are retained in ER and upon biotin treatment, they get exported from ER becoming visible in Golgi apparatus and subsequent organelles along the secretory pathway. Understanding Neuronal and Molecular Mechanisms Defining Behavioral Individuality Supervisor: HIRANO Yukinori / LIFS Student: CHO King Yi / BCB Course: UROP1000, Summer Behaviors are drastically shaped by experiences, while some aversive experience may induce behavioral alterations that can be retained for a long period. The central question is: how aversive experience is integrated into neural circuits affecting animal's behavioral repertoire? In current stage, we applied electric shock to 2202u wild type Drosophila and quantified their preference for wide or narrow arms using optimum 2-arms arenas after being shocked, as well as the retention duration of such preference. Our results show that flies generally shift towards anxiety-like behavioral phenotypes upon electric shock, which can be manifested as displaying a stronger disfavor towards narrower space or a stronger preference towards wide space. This claustrophobic-like preference could retain for at least 2 hours after shock. Biochemical Characterization of Histone Variants and Post-translationally Modified Nucleosomes Supervisor: ISHIBASHI Toyotaka / LIFS Student: TSUI Long Wai / BIBU Course: UROP1100, Fall Histones are evolutionarily highly conservative proteins that bind DNA in nucleus to form dense nucleosomes. A functional histone complex is an octamer, consisting two of each subunits H2A, H2B, H3 and H4. Variants of these subunits may deliver functions distinct from canonical histones. H2BFW is a variant of histone subunit H2B. Clinical data shows that mutation in H2BFW in male patients correlates with infertility. It is suspected that H2BFW functions to destabilize nucleosome in spermatogenesis, while the mutant form fails to do so. This study looks into the stability of H2BFW nucleosomes, and compare it to that of H2BFW mutant and canonical nucleosomes, using biochemical and biophysical techniques. This UROP report will be divided into three parts. In the first part, I will report on the work and results in the H2BFW study. In the second part, I will include progress on histone variant mouse H2A.P characterization. In the final part, I will give a brief reflection on my experience and discuss future research directions. 16 School of Science Division of Life Science Biochemical Characterization of Histone Variants and Post-translationally Modified Nucleosomes Supervisor: ISHIBASHI Toyotaka / LIFS Student: TSUI Long Wai / BIBU Course: UROP2100, Spring H2A.P is a testis-specific histone variant of H2A that is not well-understood. Although H2A.P is predicted to have a secondary structure that resembles canonical H2A, we found that human H2A.P localizes in the cytosol instead of the nucleus, while mouse H2A.P is more enriched in the cytosol than the nucleus in mammalian cells. Fluorescence recovery after photobleaching assay showed that human H2A.P is not bound to a stable structure in the cytosol. These interesting observations suggested that H2A.P probably does not bind DNA and form nucleosomes as typical histones do, and has distinct functions from canonical H2A. Domain-swap assay revealed that mouse H2A.P N-terminal and the histone-fold domain together are sufficient to exclude H2A from the nucleus and mouse H2A.P C-terminal may contribute to nucleus localization. Investigating the Effects of MEF2D on Transcription Dynamics at a Single-Molecule Level Supervisor: ISHIBASHI Toyotaka / LIFS Student: LEONG Kin Nam / BCB Course: UROP1100, Fall Previous research has shown that MEF2D (Myocyte-specific enhancer factor 2D) protein enhances polymerase transcription activity, however, a quantitative measurement of its effects has not been done. This project aims to investigate the quantitative effects of MEF2D on the dynamics of RNA polymerase transcription at a single molecule level, using optical tweezers. Results are unavailable at the time this report was written but has significant progress since the last progress report. This report focuses on the recap of the project, changes made to the experiment system, improvements that enabled progress and the current stage of the experiment. 17 School of Science Division of Life Science Investigating the Effects of MEF2D on Transcription Dynamics at a Single-Molecule Level Supervisor: ISHIBASHI Toyotaka / LIFS Student: LEONG Kin Nam / BCB Course: UROP2100, Spring MEF2D (Myocyte-specific enhancer factor 2D) is present in nucleus and mitochondria and enhances RNA polymerases transcription activity. The nucleus transcription enhancement effects by MEF2D are relatively well studied, whereas the transcription enhancing mechanism by MEF2D in mitochondria remains unknown. This project aims to investigate the effects of MEF2D on the dynamics of mitochondrial RNA polymerase (RNAPMT) transcription and further examine the effects quantitatively using quantitative sequencing techniques, in vitro transcription elongation assay, and singlemolecule optical tweezers transcription elongation assay. Comparing the WT and CRISPR-Cas9 MEF2D knockout 293T cells, we found that RNAPMT has multiple pausing sites throughout the human mitochondrial genome, and MEF2D can modulate some pausing sites. These results are confirmed by in vitro biochemical transcription elongation assay. Furthermore, to quantitatively measure its effects on RNAPMT, we have established high-spatial and temporal resolution optical tweezers transcription elongation assay. This study will allow us to quantify the effects of MEF2D on mitochondrial transcription. Investigating the Effects of MEF2D on Transcription Dynamics at a Single-Molecule Level Supervisor: ISHIBASHI Toyotaka / LIFS Student: MALLORCA Paolo Miguel Magallanes / BISC Course: UROP1000, Summer Transcription is a step that RNA polymerase uses DNA as a template to synthesize RNA. In Eukaryotic cells, genomic DNA is formed chromatin structure. It is known that condensed heterochromatin is associated to transcription repression whereas the loosen euchromatin is linked with transcription activation. The fundamental basic unit to form chromatin is histone proteins. There are core histones (H2A, H2B, H3 and H4) and linker histone (H1). Both core and linker histones have several types of histones and histone variants. It is known that some histone variants change their chromatin stability, thus RNA polymerase is much easier to transcribe through some histone variants containing chromatin. I focused to study histone variants mFM, TH2B, TH2A, and H2AL2, and how they change for the effect of transcription regulation. They are known to be expressed specifically during spermatogenesis. Their functions are not yet fully understood but are believed to play a role in histone-to-protamine transition as spermatogenic cell differentiation takes place. The aim of this project is to discover histone chaperones involved in the giving rise to these histone variants. In this UROP1000 course, we focused on creating and cloning the full recombinant DNA template construct containing a histidine tag, a sumo tag, two histone variants, and a flag tag fused together. 18 School of Science Division of Life Science Understanding the Key Elements of Nucleosome/Chromatin Dynamics Supervisor: ISHIBASHI Toyotaka / LIFS Student: THEN Claudia Regina / SSCI Course: UROP1000, Summer The nucleosome is comprised of approximately 146 base pairs of DNA wrapped around a histone octamer, consisting of two copies of each core histone H2A, H2B, H3, and H4. Despite histones being highly conserved throughout evolution, several forms of histone variants have been found. In this study, we focused on four types of variant histones, namely H2BFM, H2AL2, TH2A, and TH2B. All of the variant histones mentioned are expressed in testes, however their functions remain unknown. To understand their functions, we made several plasmid constructs containing H2BFM, H2AL2, TH2A, and TH2B for expressing recombinant proteins with the Flag tag. These proteins can be used for chromatin immunoprecipitation assay to find the interaction partners in testes. Deciphering Epigenetic Changes in Cancer Supervisor: LEUNG Danny Chi Yeu / LIFS Student: KRISHNAMOORTHY Suhas / BIBU Course: UROP2100, Fall Epigenetics refers to the study of heritable changes in gene expression without alterations in DNA sequences. In cancer cells, genetic alterations are often preceded by or accompanied by alterations in the epigenetic makeup of the cell. Differential binding of transcription factors, histone modification marks and DNA methylation states have been observed in various types of cancer cells and could provide insight into the underlying mechanism of these cancers. Aberrant epigenetic patterns can potentially serve as powerful diagnostics tools such as classification markers, risk assessment markers and sensitive detection markers. In this project, differential analysis is done using ChIP seq data of the transcription factor CTCF as well as whole genome bisulfite sequencing data from A549 lung carcinoma cells. Regions which have differential enrichment for both CTCF binding and DNA methylation are then annotated to identify regions of interest. Bisulfite sequencing is then done at one of these regions to identify the DNA methylation pattern at base pair resolution. Deciphering Epigenetic Changes in Cancer Supervisor: LEUNG Danny Chi Yeu / LIFS Student: LEE Yeung / BCB Course: UROP1100, Summer Epigenetics focuses on the chemical modifications on DNA and histones which have significant contributions towards cell diversity through dynamic regulation of gene expression. Research in this area has profound significance uncovering mechanisms in molecular biology as well as clinical application such as in the investigation of glioblastoma multiforme. This paper briefly summarizes the basic molecular biology techniques essential to epigenomic research. 19 School of Science Division of Life Science Receptor Based Drug Development from Chinese Herbal Medicine Supervisor: LI Ning / LIFS Co- supervisor: XUE Hong / LIFS Student: LEE Jiho / BIOT CHAN Teresa / BIBU Course: UROP1100, Spring UROP2100, Spring Wogonin, a member of flavone, is academically recognized for several therapeutic potentials. In particular, its anxiolytic activity by its affinity with the benzodiazepine (BZD) binding site on the γ-aminobutyric acidA (GABAA) receptor was recently discovered, making wogonin a candidate for relieving schizophrenic symptoms with fewer side effects. Extraction methods of this compound were extensively discussed in the literature, yet few were suitable for the industrial process scale-up. Dispersed Mobile-phase Countercurrent Chromatography (DMCC) provides continuous high-resolution separation for various compound types, with the potential to be applied to a large-scale wogonin purification. In this report, the extraction efficacy of wogonin from Scutellaria baicalensis Georgi with the DMCC method is to be discussed and evaluated. Receptor Based Drug Development from Chinese Herbal Medicine Supervisor: LI Ning / LIFS Co- supervisor: XUE Hong / LIFS Student: NG Chi Kit / BIBU Course: UROP1100, Spring Cancer, identified as a human complex disease by a combination of multiple gene defect and environmental factors, are regarded as one of the diseases with high fatality rate. To elucidate the genetic causes of the arise of cancer for an early analysis purpose, inter-Alu PCR sequencing is adopted to amplify the DNA sample concentration to a large extent, which can be identify the genomic sequence with a significant observation. Hence, through comparing the sequence among lung cancer patients and controls by a series of procedures, specifically BioRuptor fragmentation, gel fragment separation, end repairing, adenylation, indexation and finally next-generation sequencing, the DNA library can be constructed, and the single nucleotide variations can be identified as the fundamental for further research. Receptor Based Drug Development from Chinese Herbal Medicine Supervisor: LI Ning / LIFS Co- supervisor: XUE Hong / LIFS Student: WONG Hiu Man / BCB Course: UROP1100, Spring Alzheimer’s disease (AD) is one of the deep-rooted problems for the aging population due to the lack of potent drugs for complete treatment. To assist early diagnosis, it is of utmost importance to examine possible causes of this disease. It has been suggested Schizophrenia (SZ) could lead to higher risk of developing AD at later stages. Although recent studies have been extensively conducted to examine morphological and clinical similarities and differences between two diseases, there is no comprehensive research to analyse and compare the results of the behavioural tests and biochemical assays. This report will demonstrate the experimental results of behavioural tests on GABAA β2 subunit knockout mice to evaluate the effect of SZ on AD. 20 School of Science Division of Life Science Receptor Based Drug Development from Chinese Herbal Medicine Supervisor: LI Ning / LIFS Co- supervisor: XUE Hong / LIFS Student: LEE Jiho / BIOT Course: UROP2100, Summer Plant-based compounds are recently acknowledged as a potential therapeutic compounds. Wogonin, a member of the flavone family, is one of the academically recognized compounds which shows high anxiolytic activity substituting benzodiazepine. However, due to its structural similarity with other flavonoids — Baicalein and Oroxylin-A, the chemical isolation and purification of wogonin has faced difficulty. In this report, results from analytical HPLC with different solvents will be shared with an amendment on Dispersed Mobile-phase Countercurrent Chromatography (DMCC). Receptor Based Drug Development from Chinese Herbal Medicine Supervisor: LI Ning / LIFS Co- supervisor: XUE Hong / LIFS Student: WONG Hiu Man / BCB Course: UROP2100, Summer Aging involves progressive degeneration of bodily function and often associated with anxiety psychologically. With the rapid worldwide aging population, numerous anti-aging potential candidates have been proposed to ameliorate the age-related ailments and enhance youthful characteristics. However, most of the potential candidates are poorly investigated and have not yet reached preclinical study. Additionally, there is limited drug candidates with synergistic anti-aging and anti-anxiety effect. This report will compare the behavioural test results of mice treated with the saline and drug, which is developed by this laboratory. The anti-anxiolytic effect can be demonstrated by the behavioural tests while anti-aging effect and safety of the drug require further research investigation. Anticancer Drug Resistant Mechanisms Supervisor: LIANG Chun / LIFS Student: CHEUNG Yan Tung / SSCI Course: UROP1100, Fall Cancer is one of the top killers in human population. Therefore, there is a urgent need for the research of cancer therapeutics and mechanisms of drug resistance. To reach drug development optimization and contribute to cancer biology research, novel target drugs need to be screened. In this project, drug candidates with anticancer potential were screened for their anticancer activity and ability to inhibit DNA replication. A candidate nicknamed A-C5 was found to be a potential candidate for further investigation. 21 School of Science Division of Life Science Anticancer Drug Resistant Mechanisms Supervisor: LIANG Chun / LIFS Student: CHIU Yui Hei / BIOT Course: UROP1100, Fall M2 is a potential anti-cancer drug candidate studied in Dr Liang Chun’s laboratory. The drug candidate shows its exceptional potential and specificity in inhibiting the growth of cancer cells, particularly Hela cells. However, the long-term use of the drug exhibits a critical problem– Drug resistance. Previously, a resistance population- R8 was studied in depth through different experiments, confirming its identity and characteristics. This progress report devotes itself into studying a higher resistance population- R10, a cell line developed from Hela Cell line which has the highest drug resistance among the populations in our lab. We found out that R10 population, like R8, has a very distinctive morphological and chemically compared to the Hela cell line, we believe that the differences are responsible for the even more aggressive and invasive behaviour. On top of that, we isolated R10 single cell clones and characterized for M2 resistance. We found out that R10 clones display higher survivability and recoverability in the presence of drug compared to the parental Hela cell line. Anticancer Drug Resistant Mechanisms Supervisor: LIANG Chun / LIFS Student: HO Phei Ting / BIOT Course: UROP1100, Fall The purpose of this report is to further evaluate the properties of Chinese herbs on Insulin Resistant (IR) HepG2 cells in-vitro. The main focus is to decipher Chinese herbs that can ameliorate the effect on diabetic Type II patients by comparing the glucose uptake assay on different incubation time. The herbs used in this research are Momordica charantia, Pericarpium granati and Radix astragali. Different experimental procedures such as cell subculture, WST1 assay, Glucose Uptake Assay and Western Blot were conducted to study the underlying effect on IR cells and concluding with most significant herb using the experimental results gathered. Furthermore, the analysis of specific proteins that play specific roles in glucose uptake were further studied (GLUT-2 and IRS-1 against a reference protein: β-Actin). 22 School of Science Division of Life Science Anticancer Drug Resistant Mechanisms Supervisor: LIANG Chun / LIFS Student: GUO Yuchen / BCB CHIU Yui Hei / BIOT Course: UROP1100, Spring UROP2100, Spring M2 is a potential anti-cancer drug candidate studied in Dr Liang Chun’s laboratory. The drug candidate shows its exceptional potential and specificity in inhibiting the growth of cancer cells, particularly Hela cells. However, the long-term use of the drug exhibits a critical issue– Drug resistance. Previously, 2-8, 4-5 and 5-6 clones from the R10 population (Highest drug resistance HeLa population available in our laboratory) were studied in depth through various assays, confirming their morphology. This project devoted itself into studying another two clones; 2-9, 4-4 in comparison with the HeLa population by conducting various assays. We found that 2-9 and 4-4 exhibited similar trends as the previous R10 clones with slower growth, higher survivability after drug treatment and smaller in size. Unlike the previous clones, 2-9 and 4-4 exhibited different behaviours implying different degree of resistance to the M2 drug. Anticancer Drug Resistant Mechanisms Supervisor: LIANG Chun / LIFS Student: WOO Ho Wun Howard / BIOT Course: UROP1000, Summer The chemical changes in the cell environment alter the mechanical cues that lead to variations in biochemical signaling and inherent process such as mitosis. In the current study, we explored the process of Daughter Number Variations (DNVs) in M2 drug-resistant single cell line clones (SCCs) relative to parental HeLa. The three SCCs derived from the R10 resistant population were selected to characterize for resistance mechanism and DNV events via live-cell microscopy for 24 hours. The SCCs observed resulted in stress cell division: DNVs events (from 3 to 5 daughter cells), delayed mitosis, uneven prolonged cell size, and cell death induction. In resistant clones, the multi-daughter division was more common and exhibited amplified cell size and extended division time. Cell fusion after the division was also seen in SCCs. However, further experiments are required to understand the mechanism at the molecular level. 23 School of Science Division of Life Science Chinese Herbal Medicine for General Health Supervisor: LIANG Chun / LIFS Student: LAU Siu Wa / BCB Course: UROP1000, Summer Traditional Chinese medicine (TCM) have been widely applied in the Chinese community for thousands of years. It is also known as the oldest system of medicine in the world. In the old practice, practitioners prescribed drug based on their experience and literatures written by the predecessor. However, the performance of these herbal medicines is not well elucidated. So, this project aims to characterize herbs that can treat type 2 diabetes. We would like to investigate the performance of TCM in curing the disease. We have chosen Wolfiporia extensa (茯苓), Glycyrrhiza uralensis (甘草) and Astragalus membranaceus (黃芪) as the samples. It turned out all of them could facilitate the glucose uptake of insulin resistant cells (HepG2), hence lowering the glucose level in the culture medium. However, the difference among various concentrations was not distinct due to the uneven distribution of cells. We have also studied the toxicity of the TCM. All samples showed less toxicity than metformin, which is the first-line medication for the treatment of type 2 diabetes. Some of them could even promote the viability of cells. Chinese Herbal Medicine for General Health Supervisor: LIANG Chun / LIFS Student: WU Zehaoyu / BCB Course: UROP1100, Summer Dyslipidaemia, especially hyperlipidaemia has become known for its threat to the cardiovascular system in the recent years and various treatments are designed for this disorder. It is noticed that traditional Chinese medicine (TCM) also has the potential to treat this disorder but with a lower side effect. This study utilized HepG2 cell line to establish a hyperlipidaemia cell model to examine the preventing and relieving efficacy on hyperlipidaemia of two TCM single herbs, ChenPi and HuangQi by measuring the level of protein content, triglyceride (TG) and total cholesterol (TC) and the visualization of ORO assay, which significantly demonstrates the lipid controlling effect of two tested TCM components. DNA Replication-Initiation Proteins in Budding Yeast Supervisor: LIANG Chun / LIFS Student: SUN Chang / SSCI Course: UROP1100, Fall Noc3p is a key protein required for DNA replication licensing. It is required for the assembly of various other DNA replication-initiation proteins (DRIPs) for the initiation DNA replication. Based on previous studies in our lab, it is known that Noc3p and ORC subunits self-interact and form dimers during pre-RC formation. However, the mechanism of such self-interaction remains unclear. To further elucidate the mechanism of Noc3p self-interaction, in this study, NOC3-FRB cells were used to deplete free Noc3p from the yeast cell nucleus and determine the impact of depletion on Orc2p self-interaction. 24 School of Science Division of Life Science DNA Replication-Initiation Proteins in Budding Yeast Supervisor: LIANG Chun / LIFS Student: CHENG HE Emily Isabel / BCB GAO Xuanting Tina / BCB Course: UROP1000, Summer UROP1000, Summer Nucleolar complex associated protein 3 (NOC3) plays a crucial role in DNA replication initiation. Fully elucidating the role and function of this protein will enhance our understanding of DNA replication and lay the foundation for further application in cancerrelated treatment. It is hypothesized that phosphorylation plays an essential part in the dimerization of NOC3, Origin Recognition Complex (ORC) dimerization and pre-RC formation. To verify this hypothesis, yeast cells expressing endogenous NOC3-HA protein were used. Chromatin Binding Assays and immunoblot analysis were carried out to examine the phosphorylation status of Noc3p. The result of the immunoblotting indicates that Noc3p undergoes dephosphorylation in G1 phase and phosphorylation in S and M phases, which confirms the project hypothesis. Human Complex Disease Genomics and Bioinformatics Supervisor: LIANG Chun / LIFS Co-supervisor: XUE Hong / LIFS Student: CHAN Wing Lam / BIBU Course: UROP1100, Fall Transposons are sequences that can either move or jump within a genome. It plays a crucial role in human evolution by rising new genes through the process of domestication. Meanwhile, transposon insertion potentially disrupts the normal function of genes. Recently, some research groups discover that transposons may contribute to the formation of tumors and thus results in colorectal cancer. In view of the role of transposons, Alu-scan sequencing was performed in this study to understand the linkage of transposons and the causes of human colorectal cancer. By performing the whole-genome sequencing technique, it is aimed to study the distribution of Alu elements in genomes of patients with colorectal cancer and healthy people. Human Complex Disease Genomics and Bioinformatics Supervisor: LIANG Chun / LIFS Co-supervisor: XUE Hong / LIFS Student: IU Wing Ting / BIBU Course: UROP2100, Fall There are many proposed genetic causes for schizophrenia, a complex human disease affecting the brain. In which, the Gabrb2-origin has been studied extensively recently. Gabrb2 is the gene coding for the β2 subunit of the GABAA receptor and studies found that the Gabrb2-knockout mice model displayed schizophrenia-like behavior, which can be a potential model to study the genetic cause of schizophrenia. This project aims to study the neuroinflammation status in different ages of Gabrb2-knockout mice by measuring the amount of pro-inflammatory cytokines: interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α), by Enzyme-Linked Immunosorbent Assay (ELISA). With the data obtained, the start of neuroinflammation can be known, and the most suitable time for drug administration can be determined. 25 School of Science Division of Life Science Human Complex Disease Genomics and Bioinformatics Supervisor: LIANG Chun / LIFS Co-supervisor: XUE Hong / LIFS Student: KAN Chloe / BTGBM Course: UROP3100, Fall The GABAA receptors are pentameric protein channels that mediate neural inhibition in the central nervous system, hence they are crucial for the balance of neuronal activity. Due to their function, the receptors play a key role in neuropsychiatric disorders such as schizophrenia and epilepsy. Many psychotropic drugs target the GABAA receptors, so to better design effective drugs, the structure of GABAA receptors and the key residues for binding should be clearly identified. Hence, this study aims to investigate the structure of the key fragments of the α1, β2, and γ2 subunits, as well as the native isoform of GABAA receptors. This will help obtain a higher resolution atomic structure of the receptor. Human Complex Disease Genomics and Bioinformatics Supervisor: LIANG Chun / LIFS Co-supervisor: XUE Hong / LIFS Student: KAN Chloe / BTGBM Course: UROP4100, Spring The GABAA receptors are channel proteins that contribute to the inhibition of neurotransmitters in the central nervous system, and mediate neuronal balance. The receptors are crucial in many neuropsychiatric disorders, including schizophrenia. Drugs such as benzodiazepines also target these receptors, so the structure of the receptors are key to identifying and designing better medicine that is more effective for treating disorders. Therefore, this study aims to explore and visualize the structure of the GABAA receptors and investigate the effects of the three predominant subunits on drug binding. Hopefully, this study will lead to a deeper understanding of the atomic structure of GABAA receptors and aid in the development of neuropsychiatric drugs. Human Complex Disease Genomics and Bioinformatics Supervisor: LIANG Chun / LIFS Co-supervisor: XUE Hong / LIFS Student: HUI Cheuk Him / BIBU Course: UROP1100, Summer Lung cancer is the leading cause of cancer deaths worldwide. With the goal of improving cancer prognosis and treatment, Next Generation Sequencing (NGS) was adopted to discover the genetic causes of this disease. Prior to NGS, DNA Libraries of nine samples were prepared. Procedures include DNA extraction, Alu PCR, fragmentation, end repairing, adenylation, adapter ligation, and KOD PCR. Five to eight cycles of ultrasonic shearing succeeded in creating DNA fragments of sizes within 300 and 400bp. Five purification rounds resulted in 7.5 % recovery of desired DNA fragments. The largest decline in DNA amount occurred after KOD PCR. 99 to 213 ng of DNA remained per sample. 26 School of Science Division of Life Science Potential Cancer Drug Identification and Characterization Supervisor: LIANG Chun / LIFS Student: KWON Daye / SSCI Course: UROP1100, Fall Studies have indicated the potentials of TCM (Traditional Chinese Medicines) for treatment of insulin resistance. In the present experiment, the efficiencies in the uptake of glucose of three popular TCMs—Chinese Cinnamon, Polygonatum odoratum, Radix bardanae were compared to that of a commonly known diabetes drug, Metformin. Insulin-resistant HepG2 cells were treated with each drug solutions and were screened with addition of glucose assay kit. Absorbance at 490 nm of the treated HepG2 were measured and a glucose standard curve was plotted to obtain the absolute glucose uptake quantity for comparison. Although metformin showed the highest uptake of glucose, the results also suggested high efficiencies of glucose uptake stimulation and potentials of Polygonatum odoratum and Radix bardanae. Potential Cancer Drug Identification and Characterization Supervisor: LIANG Chun / LIFS Student: KWOK Nicholas Tze Yin / BTGBM Course: UROP1100, Spring With the proposal of minichromosome maintenance (MCM) proteins as cancer hallmarks, especially on enhanced proliferation and regulation of replicative stress, we aimed at discovering anticancer drugs via analyzing the drug efficacy towards MCM inhibition. AC5, a small molecule obtained from a preliminary screen of a chemical library, shows efficacy against MCM chromatin binding in cancer cell lines, i.e. HeLa and MCF7, and hence cytotoxic effect. It also shows selectivity, causing less damage in normal cell line MCF10A and significantly better recovery than cancer cell lines. This study shows the potential of AC5 as a novel anticancer drug candidate for later preclinical research. Potential Cancer Drug Identification and Characterization Supervisor: LIANG Chun / LIFS Student: WONG Lok Lam / BIBU Course: UROP1100, Spring Chinese herbal drugs have long been recognized for the focus on holistic physical health condition and thus have few side effects. However, it is more common to use Western oral drugs to treat diabetes nowadays. With the consideration of the increasing incidence of diabetes, there is a need to discover more efficient drugs. Chinese herbal drugs may have advantages over the traditional medical approach, in terms of efficacy and side effects. In this experiment, the glucose uptake level of insulin resistance cells is investigated after herbal drugs treatment, so as to understand the efficacy of the drugs. WST-1 is also used to test the cell viability to ensure that the drugs are not toxic to cells. 27 School of Science Division of Life Science Potential Cancer Drug Identification and Characterization Supervisor: LIANG Chun / LIFS Student: WU Zehaoyu / BCB Course: UROP1100, Spring Dyslipidaemia is one of the most significant factors that leads to the cardiovascular diseases (CVD) around the world. In China, traditional Chinese medicine (TCM) has been applied long to prevent and treat this disorder in favour of its lower adverse effects. This study utilized HepG2 cell line to establish a hyperlipidaemia cell model to examine the efficacy of two TCM single herbs, ChenPi and HuangQi through WST-1 and ORO assay. Comparing the outcome with the classic statin and fibrate medications, the efficacies of TCM are around a similar level. The potential of two tested components on lipid controlling is thus well exhibited. Potential Cancer Drug Identification and Characterization Supervisor: LIANG Chun / LIFS Student: LEE Tung Yeung / SSCI WU Chak Lam Jerome / BCB Course: UROP1000, Summer UROP1000, Summer Minichromosome Maintenance (MCM) complex is crucial for DNA replication initiation and elongation in eukaryotes and was shown to increase cell proliferation through enhancing DNA replication rate. Therefore, drugs inhibiting the assembly of MCM complex is a valid approach in curing human cancer. In this project, RPE-1 cells stably expressing MCM2- EGFP is generated and characterized aiming to set up a platform for fast and convenient screening of potential anticancer drugs. Moreover, suitable conditions for drug screening using the platform have also been tested to enable better measurement of MCM-inhibition ability of candidate drugs in preliminary studies. Molecular Regulation of Axon Regeneration Supervisor: LIU Kai / LIFS Student: CAO Jiaming / BCB Course: UROP4100, Fall The human nervous system consists of central nervous system (CNS) and peripheral nervous system (PNS). In adult mammals, PNS axons can regenerate after injury while CNS axons do not have the ability to regenerate after lesion. Understanding CNS axon regeneration is an important step that in the future could lead to treatments for spinal cord injury, stroke, ALS and also other neurodegenerative disorders. There are both extrinsic factors and intrinsic factors for the regeneration failure of adult mammalian CNS axons, and the molecular mechanisms behind axon regeneration is still under research currently. To research the molecular regulation, we manipulate genes in vivo or in vitro using mouse model and observe the regeneration extent. 28 School of Science Division of Life Science Molecular Regulation of Axon Regeneration Supervisor: LIU Kai / LIFS Student: GAO Zhouyang / BCB Course: UROP3100, Spring After injury, most axons in the central nervous system (CNS) of adult mammalians fail to regenerate while axons in the peripheral nervous system (PNS) can regenerate robustly. Studying the underlying mechanism of PNS axon regeneration can provide insights into treating PNS injury, and the knowledge may be transferred to CNS injury. Conditioning lesion is a well-established strategy to enhance the regeneration capacity of dorsal root ganglion neurons (DRGs). So far, whether the immune system is involved in this process is not well-described. We are interested in the interferon (IFN) family, a featuring group of proteins involved in the immune response. We found that IFN-γ is involved in this process and regulating IFN-γ downstream signaling factor can modulate DRG axon growth. Molecular Regulation of Axon Regeneration Supervisor: LIU Kai / LIFS Student: LIU Jingyi / BIBU Course: UROP1000, Summer The failure of axon regeneration in the adult mammalian central nervous system (CNS) are attributed to two properties of the adult CNS, the inhibitory extrinsic environment, and a diminished intrinsic regenerative capacity of mature CNS neurons. The project is to identify and study the novel genes involved in axon regeneration, particularly through intrinsic regulation. The axon regeneration can be affected by a number of genes and the expression level of those genes can be promoted or inhibited by a number of compounds. I have tested two batched of compounds and selected effective ones and repeated once using PC12 cell lines, and next, in DRG cell lines. The Rules of Packaging Fat in Cells Supervisor: MAK Ho Yi / LIFS Student: MA Chengkun / BCB Course: UROP1100, Fall tmem-120 is a poorly characterized gene in C. elegans. Through RNA-sequencing, highly differentially expressed genes between wild type and tmem-120 loss-of-function mutant worms have been identified. This UROP project aims at producing transgenic worms for validating the RNA sequencing results in vivo. The green fluorescent protein (GFP) will be expressed under the control of regulatory sequences, including promoters and 3’ untranslated regions (3’-UTR) of five genes, whose expression was strongly responsive to the loss of tmem-120 function. The transgenic worms can further be used as reporters of TMEM-120 activity in genetic screens for modulators of TMEM-120 function. During this semester, plasmids for generating transgenic worms have been constructed. They will be micro-injected into worms in order to yield singlecopy transgenes in the future. 29 School of Science Division of Life Science The Rules of Packaging Fat in Cells Supervisor: MAK Ho Yi / LIFS Student: MA Chengkun / BCB Course: UROP2100, Spring tmem-120 is a gene identified in a genetic screen for lipid droplet expansion suppressors. It was shown that tmem-120 loss-of-function mutation could affect fatty acid accumulation to suppress lipid droplet expansion. The first part of this UROP project aims to produce transgenic worms that carry GFP reporters responsive to TMEM-120 activity, which is a continuation of the previous UROP project. The second part of this project intends to test a newly improved technique, the auxin-inducible degron 2 system, in C. elegans. Ideally, this technique allows efficient and reversible protein degradation in response to the plant hormone auxin. An approach similar to the first part is adopted to generate transgenic animals for testing. Characterization of Novel Cell Cycle Regulators in Cancer Cells Supervisor: POON Randy Yat Choi / LIFS Student: CHEUNG Lo Sha / BICH Course: UROP1100, Fall As a member of MYC family, MYCN gene amplification has been shown in multiple human tumors, including a severe form of neuroblastoma. In this research project, by deploying CRISPR-Cas9 technology, different sequences are targeted, causing specific sites of double-strand break (DSB) on MYCN gene, inducing deactivation of MYCN genes. Subsequently, cell cycle stages were examined. A CRISPR targets exon 3’s open reading frame of MYCN was found to be able to affect cell cycle in HtTA1 N-Myc #15 cells. Since many cancers are related to abnormal cell cycle, the study of MYCN related pathway can provide insights for therapeutic development targeting MYCN gene. CRISPR/Cas9 Analysis of Essential Genes Supervisor: POON Randy Yat Choi / LIFS Student: RAG Adwika / SSCI Course: UROP1000, Summer Knockout of genes can be led to a more molecular based and specific treatment for cancer patients. One gene suggested is SPC24. Using auxin-inducible degron (AID) system to conditionally degrade necessary proteins, for which doxycycline and IAA are added to these cells to observe if they undergo mitotic arrest and cell death. Mainly done through immunoblotting and mitotic index calculations, it was observed that some degree of apoptosis was observed but it was not as comparable as other prevalent chemotherapy drugs such as Taxol or Nocodazole (Noc). Furthermore, there was an attempt to see the best dosage of these chemotherapy drugs due to their high toxic side effects, but after serial dilutions, the mitotic arrest and cell death observed through immunoblotting showed that these dilutions are not enough to cause complete death and have reduced efficacy. 30 School of Science Division of Life Science CRISPR/Cas9 Analysis of Essential Genes Supervisor: POON Randy Yat Choi / LIFS Student: SUN Run / BCB Course: UROP1000, Summer Apoptosis is the process of programmed cell death which is critical for cells to survive. Normal cells use this mechanism to clean the body of cells that are irreparable to prevent cancer cells. However, abnormal cells such as cancer cells have deficiency in this function and may grow without control. So, it is indispensable to study the mechanism behind the process of apoptosis and unveil the essential gene related to it. In this study, we try to uncover the relationship between some relevant genes and MARCH5, a known E3 ubiquitin-protein ligase localized in the mitochondrial outer membrane, and the pathway of how the deficiency of MARCH5 can lead to mitotic cell death in cancer cells. CRISPR/Cas9 Analysis of Essential Genes Supervisor: POON Randy Yat Choi / LIFS Student: ZHOU Siyu / BCB Course: UROP1100, Summer Hypusination is a unique posttranslational modification occurring only on one specific lysine residue of the universal translation factor, eIF5A. The pathway consists of two steps, the first is the processing of eIF5A precursor and spermidine to form deoxyhypusine, which is catalyzed by deoxyhypusine synthase (DHPS), and the second step is completed by deoxyhypusine hydroxylase (DOHH), producing hypusinated, mature eIF5A. While eIF5A, DHPS and DOHH are all essential for cell proliferation, knock out of DHPS causes a band shift of DOHH in Western Blot as observed by Marine Ma in our laboratory (2021). To identify the reason for such change, a recombinant FLAG-tagged DOHH construct was produced, and multiple approaches were adopted to reproduce the result. Targeting Mitotic Regulators in Cancer Cells for Potential Treatment Supervisor: POON Randy Yat Choi / LIFS Student: LIAO Man Kit / CHEM Course: UROP1100, Spring The Spindle Assembly Checkpoint (SAC) and Mitotic Checkpoint Complex (MCC) are important components during mitotic arrest. In the absence of microtubule attachment to kinetochore, a cascade of signal leads to activation of SAC and then the MCC. Inhibition of the Anaphase-Promoting Complex (APC/C) by MCC enables the cells to stay in mitosis. Previous studies suggested that MAD1 is a key component of the SAC and plays a crucial role in this activation of the MCC. Using a conditional system to deplete MAD1, however, indicated that MCC complex can still form and cells can still be arrested in mitosis in the absence of MAD1. These results suggested that MAD1 may not be as crucial for the SAC as once thought. 31 School of Science Division of Life Science Targeting Mitotic Regulators in Cancer Cells for Potential Treatment Supervisor: POON Randy Yat Choi / LIFS Student: CHING Chi Yin / SSCI Course: UROP1100, Summer It has been found by our laboratory that Programmed death-ligand 1 (PD-L1) express a second band (other than the normal band show in interphase) in Western blot during mitosis. We suspect that PD-L1 may possess a certain function in the mitotic state. We would like to investigate the identity of the second band by transfecting various modified clones of the PD-L1 gene. Prior to this, we have to confirm that the transfected PD-L1 shows a similar pattern as the endogenous one. This study mainly focuses on evaluating the potential of HeLa being the model for future experiments. Unfortunately, the result indicates that transfected HeLa shows a different PD-L1 expression pattern than the normal cell line. Therefore, it is not a suitable option for the future experiments. Targeting Mitotic Regulators in Cancer Cells for Potential Treatment Supervisor: POON Randy Yat Choi / LIFS Student: LIAO Man Kit / CHEM Course: UROP2100, Summer The Spindle Assembly Checkpoint (SAC) and Mitotic Checkpoint Complex (MCC) are important components during mitotic arrest. In the absence of microtubule attachment to kinetochore, a cascade of signal leads to activation of SAC and then the MCC. Inhibition of the Anaphase-Promoting Complex (APC/C) by MCC enables the cells to stay in mitosis. Previous studies suggested that MAD1 and BUB1 are key components of the SAC and plays a crucial role in this activation of the MCC. Using a conditional system to deplete MAD1 and BUB1 respectively, however, indicated that MCC complex can still form and cells can still be arrested in mitosis for a shorter period of time in the absence of MAD1 or BUB1. These results suggested that both MAD1 and BUB1 may not be as crucial for the SAC as once thought. Cloning and Characterization of Novel Cytoskeletal Regulators Supervisor: QI Robert Zhong / LIFS Student: PARK Joo Hyoung / BIOT Course: UROP1100, Fall Microtubules are a type of cytoskeletal network which plays an important role in the transportation of cellular materials and structural support of the cell. These microtubules are generated at sites known as the microtubule organizing centers (MTOCs), with its nucleation beginning at the γ-tubulin ring complexes (γ-TuRC). The γ-TuRC is a built with γ-tubulin proteins and various γ-tubulin complex proteins (GCP). This report covers the cloning of C-terminal EGFP tagged GCPs into pEGFP-N3 vectors and transformation into bacteria, with the final goal being to compare C-terminal tagged GCPs to N-terminal tagged GCPs in order to determine if it is the GCPs or tagged EGFPs cause interference with regular γ-TuRC function. 32 School of Science Division of Life Science Cloning and Characterization of Novel Cytoskeletal Regulators Supervisor: QI Robert Zhong / LIFS Student: PARK Joo Hyoung / BIOT Course: UROP2100, Spring CRISPR is a novel genetic engineering technology widely applied to many areas of biochemistry and cell biology, originating from the naturally occurring adaptive immunity response system in bacteria. Its uses involve knockout of genes, activation/repression of genes, purification of DNA, precise editing of DNA and RNA, and more. This report discusses the use of this technique with applied assays such as microtubule regrowth assay and FACS assay to be able to identify and study in details the genes responsible for the nucleation of microtubules at MTOCs, with additional insight into how exactly CRISPR could be applied and focusing particularly on plasmids including psPAX-2, pMD-2G, pBABE-ZEO. Cloning and Characterization of Novel Cytoskeletal Regulators Supervisor: QI Robert Zhong / LIFS Student: CATIELLO Carlo / BIOT Course: UROP1000, Summer The γ-tubulin ring complex (γ-TuRC) is a protein complex composed of γ-tubulin small complexes (γTuSC) and multiple γ-tubulin complex proteins (GCP) that regulates microtubule nucleation by providing control over microtubule growth. Neural precursor cell expressed developmentally down-regulated protein 1 (NEDD1) is an important component of the centrosome that directly interacts with γ-tubulin and targets it to the chromosome. In this experiment, conventional molecular cloning methods were used to clone NEDD1 579-667 to utilize it as a bait to pull down γ-TuRC-bound NEDD1 from the centrosomes. A mutated version of the protein, NEDD1 579-667 H615F, which have reduced binding ability to γ-tubulin was also cloned to be used as a control. Confirmation of the experiment was done by double digestion as well as DNA sequencing. Investigation of Novel Mechanism Underlying Microtubule Organization Supervisor: QI Robert Zhong / LIFS Student: AU Chun Kei / CIVL Course: UROP1000, Summer The γ-tubulin plays an important role in microtubule organization. Under normal situations, microtubule is made up of both lateral and longitudinal γ-tubulin interactions by polymerization. In this research study, our goal is to detect the lateral binding acid also known as γ-tubulin-4A-GFP. It is known that the mutant protein γ-tubulin-4A will disrupt the longitudinal γ-tubulin interaction. Therefore, when we construct the γ-tubulin-4A mutant, it will disrupt the γ-tubulin/𝛼-tubulin and γ-tubulin/γ-tubulin longitudinal interaction leaving only the lateral γ-tubulin interactions. So, the detection of lateral γ-tubulin interactions is crucial for investigating the γ-tubulin-4A mutant protein. To aid our investigation on the mutant protein and its construct, we need to carry out several experimental procedures. 33 School of Science Division of Life Science Standardization of Traditional Chinese Medicines Supervisor: TSIM Karl Wah Keung / LIFS Student: ABDRASSILOVA Kamila / BIBU Course: UROP1100, Spring Ameliorating skin conditions, such as atopic dermatitis (AD) and psoriasis, are linked to chronic inflammation. Edible bird’s nest (EBN) has demonstrated anti-inflammatory activity but the mechanisms of action are still investigated. Here, we review signalling pathways and proteins during inflammation that potentially could be targeted by EBN’s anti-inflammatory action. Moreover, the cellular synthesis of filaggrin, a protein being downregulated in AD, has been shown to be regulated by GATA3, a transcriptional factor. Here, we demonstrate results from RT-PCR, where EBN extract, EBN digest and N-acetylneuraminic acid (NANA) lead to significant increase in expression of GATA3 mRNA. Additionally, we suggest that the action of EBN having PPARα or PPARγ as modulator is worth to be investigated. Understanding the mechanism behind anti-inflammatory properties of EBN can be applied to the development of topical creams in skin diseases. Standardization of Traditional Chinese Medicines Supervisor: TSIM Karl Wah Keung / LIFS Student: CHOI You Jin / BIOT Course: UROP1100, Spring Human skins, eyes, and hairs exist in a wide spectrum of colors and gradients due to the production of certain pigments known as melanin, which provide a defence mechanism against UV radiation and reactive oxygen species (ROS). Melanin pigments are produced and accumulated in melanosomes, a particular organelle within melanocytes, through the process called melanogenesis. A diversity of literature reviews demonstrates that melanosomes are being transported from melanocytes to a neighboring cell, named keratinocytes, to perform various functions, such as proliferation, differentiation, dendritogenesis, and melanogenesis. However, an abnormal increase in melanin production may induce medical problems, therefore demanding scientists are aiming to understand factors causing melanogenesis, in order to produce therapeutic solutions. Standardization of Traditional Chinese Medicines Supervisor: TSIM Karl Wah Keung / LIFS Student: WONG Tsz Yan / BIBU Course: UROP1100, Spring Cannabidiol (CBD) prevalent cannabis derivatives are increasingly popular and widely available on the market as replacement of THC, tobacco substitutes or therapeutics for various health conditions. The most two common ways to take CBD are oral and topical in which delivery methods include capsules, tinctures, balms, creams and more. In this project, CBD volatility in traditional Chinese medicine (TCM) essential oil is evaluated, which aims to test the feasibility of pulmonary administration in intaking CBD apart from traditional methods. This report devotes itself to utilizing High Performance Liquid Chromatography (HPLC) system coupled with Triple Quadrupole LC/MS to study the concentration of CBD in essential oil after a designated period of time, and the data could be analyzed to determine the best TCM essential oil in evaporating CBD during the administration. 34 School of Science Division of Life Science Computational Study of Long Noncoding RNAs in Cancer Supervisor: WANG Jiguang / LIFS Student: LEE Jooran / BIEN Course: UROP1100, Fall The identification of the novel lncRNA in glioma was the primary focus of the research, and several mistakes were identified at the beginning of this semester while revising the previous work. An incomplete run of the FLORA, and several coding mistakes produced the wrong file with some inconsistent parts. To produce a correct file, I firstly used FLORA to generate the count matrix, followed by the FPKM matrix by normalizing the count data using the IDH samples (wildtype + mutant). There were total 462 IDH samples, and 301 were IDH-wildtype patient samples, and the rest 152 samples were IDH-mutant ones among them. Then, I focused on the IDH wildtype (wt) samples for the downstream analysis of the TMZ-treated patients, and several differentially expressed lncRNAs between recurrent and initial glioma were identified. Computational Study of Long Noncoding RNAs in Cancer Supervisor: WANG Jiguang / LIFS Student: WU Yurong / CHEM Course: UROP1100, Fall Cancer is one of the deadliest diseases in the world. Glioblastoma is a type of brain cancer that recurs easily even if the patient is treated in best effort. In this project, the focus is the analysis of long non-coding RNAs (lncRNAs) of glioblastoma with computational methods such as Python and R, and statistics using data of expression levels of lncRNAs in initial and recurrent tumors from a group of 305 patients (denoted as from G001 to G305). Using statistics, we can find out the lncRNAs that are the most upregulated and downregulated in recurrent tumors. These are the important genes that is worth further analysis including the investigation of their secondary and tertiary structures of the transcripts. Computational Study of Long Noncoding RNAs in Cancer Supervisor: WANG Jiguang / LIFS Student: WU Yurong / CHEM Course: UROP2100, Spring Cancer is a deadly incurable disease. Many scientists are working on it, hoping to discover new methods of treatment. Despite the large number of researches conducted, no effective treatments can be developed. In the light of this, we are using long non-coding RNAs (lncRNAs) as a new direction to analyze cancers. In this project, we use glioblastoma as our first example. Our objective is to find out the mechanisms of various characteristics of cancer such as proliferation. By using statistics and bioinformatics, a large number of genes were analyzed in various aspects including expression levels in initial and recurrent tumors, and conservation scores. These results can later provide us some clues to the real mechanisms of cancers. 35 School of Science Division of Life Science Computational Study of Long Noncoding RNAs in Cancer Supervisor: WANG Jiguang / LIFS Student: WU Yurong / CHEM Course: UROP3100, Summer Cancer is a deadly, incurable disease. Although many researchers are working on the problem, there is no breakthrough in the development of treatment of the disease. We are using long noncoding RNAs (lncRNAs) as a new approach to tackle the problem. In the past, scientists once thought that lncRNAs had little or no biological function, but later it was discovered that they are involved in many cellular activities, so this will be a novel direction for studying the mechanism of cancers. We used glioblastoma as an example and study the lncRNAs expressed. Then, we will try to study other kinds of tumors using lncRNAs. We hope that mechanisms of cancers can be found so that we can develop new treatments. The Application of Big Data Technologies in Precision Cancer Medicine Supervisor: WANG Jiguang / LIFS Student: CHOI Dawon / BIOT Course: UROP1100, Spring UROP2100, Summer Conducting cellular deconvolution on tumor tissues is crucial for understanding the cause and mechanism of cancer, and for identifying appropriate treatments for the patients. CIBERSORTx is one analytical tool that takes scRNA-seq data and conducts a deconvolution analysis by providing the proportion of different cell types in a mixture. In this project, CIBERSORTx was used to analyze tissue samples from GBM patients. To benchmark the various parameters provided by CIBERSORTx and evaluate the deconvolution analysis of GBM tissue samples using this tool, the procedure was repeated on two different datasets. Additional datasets will be included in future analysis to validate the robustness of the method, and the pipeline will then be applied in analyzing immune cell types in glioma evolution. The Application of Big Data Technologies in Precision Cancer Medicine Supervisor: WANG Jiguang / LIFS Student: SHAO Zhihao / BCB Course: UROP1100, Spring Chromosome instability (CIN) is a type of genomic instability in which the rate that chromosomes or large portions of chromosomes change increase. As one of the most common causes of aneuploidy and chromosome rearrangement, CIN has been reported to result in a high burden of genomic aberrations such as loss of heterogeneity (LOH) and even drug resistance. With the emergence of copy number variation (CNV) profiles based next-generation sequencing (NGS) results, CIN characteristics of one tumor sample could be analyzed and estimated in the absence of single cell data. In this report, several quantitative measurements of CIN based on CNV profiles are proposed and performed on glioblastoma multiform (GBM) samples to validate their feasibility. 36 School of Science Division of Life Science The Application of Big Data Technologies in Precision Cancer Medicine Supervisor: WANG Jiguang / LIFS Student: LIU Ran / SSCI Course: UROP1000, Summer Whole genome sequencing is the source of data for calculating the evolutionary tree. However, this method has its systematic error. Despite the error can be reduced by improving the quality of sequencing, using high quality sequencing on every sample is quite time and money consuming. Here we create a model to simulate the measuring process and a C++ program to compute the error. This program allows us to estimate the error, unravel the factors affecting the error, infer the real data from the WGS measured data, choose an appropriate sequencing quality, test the validation of this model and hence get a more accurate evolutionary tree. The Application of Big Data Technologies in Precision Cancer Medicine Supervisor: WANG Jiguang / LIFS Student: SHAO Zhihao / BCB Course: UROP2100, Summer Glioblastoma is the most deleterious brain tumor, whereas the contemporary research into this malignant form of glioma is exceedingly limited to the Caucasian race. By sequencing a East Asian-specific cohort consisting of IDH-WT primary glioblastoma patients, we compared the clinical, genomic, and transcriptomic features of the Asian cohort with TCGA Caucasian cohort. Of note, the Asian cohort exhibited significantly better survival. SCNV analysis revealed that patients from the Asian cohort had considerably lower levels of EGFR amplification and PTEN deletion. Such racial difference in EGFR amplification level remained significant in men but diminished in women. This study illuminated the disparities in SCNV characteristics between the East Asian and Caucasian GBM cohorts and examined their correlation with distinct attributes. Study of Blood Cell Development using Zebrafish Model Supervisor: WEN Zilong / LIFS Student: FAN Yining / BCB Course: UROP1100, Fall Our collaborators identified a novel somatic gain-of-function mutation in MAP3K3 associated with Cerebral cavernoma malformation (CCM), a type of vascular malformation characterized by vein and capillary dilation and subsequent hemorrhage. Importantly, they found strong association between MAP3K3 mutation and Type II and Type III CCMs7. Here we generated zebrafish disease models of MAP3K3(MUT) sporadic CCMs in order to find out the molecular mechanisms of how MAP3K3 mutation contributes to CCM lesions. We first demonstrated that compared with zebrafish injected with hMEKK3(WT) coding construct, zebrafish injected with hMEKK3(MUT) coding construct showed more frequent hemorrhage which is a feature of CCM lesions. More experiments will be carried out afterwards and finally we will conduct drug screening utilizing zebrafish models, hoping to find out a non-invasive cure to CCMs. 37 School of Science Division of Life Science Study of Blood Cell Development using Zebrafish Model Supervisor: WEN Zilong / LIFS Student: LI Yunbo / BCB Course: UROP3100, Fall Metaphocytes are non-hematopoiesis-derived myeloid-like cell populations found in zebrafish epidermis, gill, and intestine. How metaphocytes adopt a myeloid-like cell fate becomes an intriguing question. To explore metaphocyte development, we investigated the maintenance of metaphocytes by studying metaphocyte progenitorcells (MPCs) recently found in zebrafish epidermis. MPCs are small and round cells, which are different from the ramified morphology of metaphocytes. RNA sequencing analysis further showed that MPCs do not express macrophage marker mpeg1 but express metaphocyte marker mstf1 and some proliferation-related genes. To investigate MPCs, we selected a candidate MPC marker gene, anxa2al, and synthesized antisense anxa2al in situ hybridization probe to test its labeling specificity. The double transgenic line Tg(mstf1:CreER;mpeg1:LoxP-DsRedx-LoxP-GFP) was also generated for MPC-metaphocyte lineage tracing study. Study of Blood Cell Development using Zebrafish Model Supervisor: WEN Zilong / LIFS Student: FAN Yining / BCB Course: UROP2100, Spring In order to find out molecular mechanisms of cerebral cavernoma malformation (CCM) associated with MAP3K3-I441M, a recently identified mutation, we generated several zebrafish disease models. Among these models, Bacteria Artificial Chromosomes (BACs) transgenic lines are necessary because it is able to mimic the endogenous expression patterns of the MAP3K3 gene. This semester we tried to generate BAC-transgenic lines and all cassettes (including itol2 and target cassette) have already been electroporated into DH10B E.coli. Further experiments for selecting and confirming correct colonies will be carried out in the future. Study of Blood Cell Development using Zebrafish Model Supervisor: WEN Zilong / LIFS Student: LI Zhuying / SSCI Course: UROP1100, Summer Aggregation of TDP-43 in neurons has been a pathological hallmark of amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD). To study the role of TDP-43 aggregation in vivo, we generated transgenic zebrafish overexpressing either wild-type TDP-43z or a mutated form with a M337V substitution. These overexpression lines showed neural abnormalities, including cytoplasmic mislocalization of TDP-43z and fragmentation of TDP-43z overexpressing neurons. Furthermore, we observed microglia uptaking the TDP-43z positive debris. Altogether, we established TDP-43 overexpression models in zebrafish which will serve as a powerful tool to study the underlying mechanisms of neurodegenerative diseases and neuron-immune cell interactions in pathological conditions. 38 School of Science Division of Life Science Calcareous Dinoflagellates Supervisor: WONG Joseph Tin Yum / LIFS Student: MONG Sze Hang / BIOT Course: UROP1000, Summer Being one of the calcareous dinoflagellates, Thoracosphaera heimii works as a major sink of atmospheric carbon dioxide in the world’s oceans (Elbrachter et al., 2008). Moreover, after being found in fossil records from the Cretaceous onwards, T. heimii. can be one of the Paleocene-graphic proxy archives for the reconstruction of the ancient oceanic environment and a better understanding of climate changes (Karwath et al., 2000). Yet, little is known for this species, and thus, it is worth researching the importance behind T. heimii.. With this notion, this report would like to give a fundamental understanding regarding the life cycle of T. heimii. as well as various features of it, including distribution of photosynthetic pigments and crystallization on the cell wall, through observation by light microscope and polarizing light microscope. The Intein Approach for Efficient Production and Purification of Recombinant Proteins in Escherichia Coli Supervisor: WONG Wan Keung / LIFS Student: YUK Tin Cheung / BTGBM Course: UROP1100, Fall Basic Fibroblast Growth Factors is a protein that was found to have versatile applications in the medical stream, including the wound and bone healing, the treatment for myocardial infection. Studies even revealed its potential in stimulating the angiogenesis and neuron extension. Previously, bEGF were extracted from natural sources including placenta, carpus luteum, yet the shortcomings associated with natural extraction were the small quantities and low yield of bEGF obtained. Therefore, scientists have turned to genetically modifying bacteria like E.coli for generating higher quantity of desirable bEGF products. Thanks to the previous researches, a comprehensive method has been developed to isolated the desired bEFG proteins from the cluster of fusion proteins, protein aggregates and other variants from the bacterial expression. In this report, methodologies and results will be presented to cover my experience adopting the technique for extracting the bEGF proteins from the E.Coli cells. G Protein-Coupled Receptors in Cell Signaling Supervisor: WONG Yung Hou / LIFS Student: LIU Zhiyin / BCB Course: UROP1100, Summer Melatonin receptors in mammals termed MT1 and MT2 are members in G protein-coupled receptors(GPCRs) which regulate circadian rhythm and they can form homodimers and heterodimers during signaling. Tango is an assay applied to detect GPCR activation for agonist ligands by the recruitment of β-arrestin which is G protein-indepedent and release of transcription factor (tTA) due to site-specific cleavage. It transfers the instant combination to a stable and visible luciferase gene expression and amplifies the small input signals, which provides us a novel way to detect dimerization of GPCRs. In this study, we propose that bioluminescence can be detected due to dimerization of melatonin receptors by replacing the β-arrestin with MT1/MT2 and modified Tango construct with MT1/MT2 and TEV protease are built. Similar construct modification for other GPCRs can be applied to further research on determination of dimerization. 39 School of Science Division of Life Science 40 G Protein-Coupled Receptors in Cell Signaling Supervisor: WONG Yung Hou / LIFS Student: CHOY Hiu Ching / BIBU Course: UROP1100, Spring Nm23 protein was the first metastasis suppressor being discovered. Nm23-H1/H2 exhibit multifunction with enzymatic activities, protein-protein interactions and DNA interactions. Studies have suggested that transcriptional targets of Nm23-H1/H2 include MMP2, EDG2, MYC and PDGF-A. PDGF-A, MMP2 and EDG2 are downregulated while MYC is upregulated by Nm23-H1/H2. This study demonstrates that Nm23-H1/H2 may indeed suppress MMP-2 transcription in 1321N1 cells and activate MYC transcription in MDA-MB-231 cells. Interestingly, Nm23- H1/H2 activated EDG2 transcription and suppressed MYC transcription in 1321N1 cells. However, the study failed to demonstrate PDGF-A gene suppression by Nm23-H1/H2. The results suggested that transcriptional regulations of cancer-associated genes may contribute to metastasis suppression function of Nm23-H1/H2, and the regulations may vary in different types of cancers. Bioinformatic Analysis of Single-Cell Genomic Data Supervisor: WU Angela Ruohao / LIFS Student: CHENANANPORN Pansit / BIEN Course: UROP1000, Summer This abstract summarizes the progress of sequencing reads using different assemblers. A sample assembly of the SARS-CoV-2 genome was performed using SPAdes. To replicate short reads produced by a sequencing machine, a complete assembly was fragmented into segments with sizes of 100 to 150 base pairs. Formatted into a FASTA file, these fragments were then inputted into SPAdes. The UCSC Blat Search Genome tool showed that the assembly was complete with a few identity mismatches. An assembly of a local sea urchin species was also performed. BUSCO was used to assess the genome by comparing its evolutionary similarity with an existing dataset (metazoa_odb10). However, there was a high percentage of duplicated orthologs. Thus, the next step, which is still currently worked on, is to remove overlaps in the assembly. Additionally, an assembly of the sea urchin data using SPAdes will be performed and compared with that of using Hifiasm. School of Science Division of Life Science Big Data: Bioinformatic Analysis of Single-Cell Genomic Data Supervisor: WU Angela Ruohao / LIFS Student: WU Ka Hei / GBUS Course: UROP1100, Summer As the global population continues to age, age-related neurodegenerative disorders are predicted to increase to 135 million cases worldwide by 2050, overtaking cancer as the second leading cause of death after cardiovascular disease (Gammon, 2014). In this study, we aim to create a singlenucleus RNA seq (snRNA-seq) library to facilitate the study of neurodegenerative disorders. We focus on analyzing the gene expression in three regions of the mouse brain – cortex, hippocampus, and cerebellum, which we will use at a later stage in this study to identify any aging-related molecular and cellular changes in the brain by comparing mice of 3-months vs. 17-months of age. This progress report will focus on featuring current annotated brain regions and account for the method used to derive these annotations. Genetic Studies on Stem Cell Regulation Supervisor: XIE Ting / LIFS Co-supervisor: TU Renjun / LIFS Student: LEE Man Ping / SSCI Course: UROP1000, Summer In the Drosophila ovary, germline stem cell (GSC) development is controlled both intrinsically and extrinsically. The niche, which provides special microenvironments for GSC development. RNA-binding proteins (RBPs) are critical in controlling gene expression through regulating RNA stability, decay, splicing or translation. However, the functions and mechanisms of RNA binding proteins in controlling germline stem cell development remain unclear. Therefore, identifying the RNA binding proteins essential for germline stem cells intrinsically and in the differentiation niche helps to understand how these genes control the development of GSC. By using genetic screening, we found that knocking down snRNP-U1-70K, clp or srrm1 in the differentiation niche with c587-Gal4 causes severe ovary development defects, differentiation defects. We also identified CG7903, pAbp, Imp and FNE are intrinsically required for GSC self-renewal. In the future studies, we will further how these RBPs control GSC self-renewal and differentiation molecularly. 41 School of Science Division of Life Science Genetic Studies on Stem Cell Regulation Supervisor: XIE Ting / LIFS Co-supervisor: TU Renjun / LIFS Student: LEUNG Sin Ho Henry / BIBU Course: UROP1000, Summer Although 5-hydroxymethylcytosine (5hmC) has been discovered in about 1952, the functions and mechanisms are yet to be known. Ten-eleven-translocation (TET) proteins function to convert 5-Methylcytosine (5mC) to 5hmC in both DNAs and RNAs. It has been reported that very low level 5mC/5hmC modified DNAs could be detected in Drosophila, therefore, Drosophila has unique advantage to study RNA 5mC/5hmC modifications. Our preliminary research found that overexpressing Tet in germline or in the niche both significantly disrupt germline stem cell (GSC) development, indicating the balance between 5mC and 5hmC in RNAs are very important. To identify the molecular partners of Tet, we knocked down many genes in Tet overexpression ovaries in this genetic screening study. We found that knockdown of genes encoding RNA 5mc modification readers (lin-28, CG9705), RNA splicing factors (mod, bru2, snRNP-U1-70K), or RNA binding proteins (C7903, Imp, Rb97D, pAbp) can partly rescue Tet overexpression caused ovary developmental defects. These results indicate the potential molecular mechanisms that how Tet involved in regulating GSC development. Genetic Studies on Stem Cell Regulation Supervisor: XIE Ting / LIFS Co-supervisor: TU Renjun / LIFS Student: LIN Muyun / BCB Course: UROP1000, Summer RNA 5mc (5-methylcytosine) is a type of RNA modification that is found in a wide variety of species. Generally, eukaryotic methyltransferases in the NSUN family and the DNA methyltransferase homologue DNMT2 catalyze most of these C5 methylation reactions. In Drosophila, it is predicted to have six RNA 5mC methyltransferases, including Nsun1, Nsun2, Nsun4, Nsun5, Nsun6 and Mt2. In this project target genes are knocked down with the use of transgenic crosses. Specifically, the Gal4-UAS system is used to express RNAi to knock down target genes in niche cells and germ cells by using nos-Gal4 and c587-Gal4, respectively. Interestingly, Nsun1 and Nsun2 double knockdown in the niche decreases germline stem cell (GSC) but increases GSC daughter cell numbers, indicating their complimentary roles in controlling GSC self-renewal and differentiation in the niche. Moreover, Nsun1 knockdown in germline completely loss GSC. In the future study, we will carefully analyse the functions of other methyltransferases in the regulation of GSC development. 42 School of Science Division of Life Science Genetic Studies on Stem Cell Regulation Supervisor: XIE Ting / LIFS Co-supervisor: TU Renjun / LIFS Student: TANG Xiaohan / SSCI Course: UROP1100, Summer In the Drosophila ovary, germline stem cells (GSCs) continuously proliferate and differentiate to enable constant egg generation. GSC self-renewal and differentiation are maintained by a variety of intrinsic and niche signals in the germarium. As one of the critical signaling pathways in animal cells, the cAMP/Pka pathway is involved in the regulation of many important growth and development processes. However, the functions of cAMP/PKA pathway in controlling GSC development remain unclear. In this study, by knocking down Drosophila cAMP/Pka pathway components in germline and the niche, respectively, we discovered that the lack of cAMP generation in both somatic cells and germline cells leads to defective GSC development. We found that, Pka-R2 knockdown completely loss GSC and its progeny, indicating cAMP/PKA pathway may involve in maintaining GSC. Interestingly, knockdown of rut, the cyclase that is responsible for synthesis of cAMP, in the niche cell also disrupts GSC development. In the future study, we will systematically analyze all cAMP/PKA components in the niche and germline to further verify our findings, and investigate the molecular mechanisms that how cAMP/PKA pathway controls germline development. 43 School of Science Department of Mathematics Department of Mathematics Statistical Analysis in Portfolio Construction Supervisor: CHEN Kani / MATH Student: SONG Wenxin / DSCT Course: UROP1100, Spring Stable coins are a new kind of cryptocurrency. They are backed by some tangible assets and are designed to obtain price stability; usually, 1:1 pegged to some widely used real-world currencies like US Dollar, Bound or Korean. It now plays a significant role in decentralized finance. This project focus on some popular stable coins and data analysis about them. This project first investigates some widely used coins like Dai, MKR, and their mechanism. Then we obtain some relevant data of some coins to do explanatory analysis to see their properties and characteristics. Further, we explore the possibility of arbitrage through stable coins. Statistical Analysis in Portfolio Construction Supervisor: CHEN Kani / MATH Student: NAN Xi / DSCT Course: UROP1100, Summer Noncustodial automated market makers such as Uniswap serve as an on-chain system of smart contracts on the Ethrereum blockchain, implementing an automated liquidity protocol. The development of Uniswap Version 3 provides increased capital efficiency and fine-tuned control to liquidity provides, improves the accuracy and convenience of the price oracle, and has a more flexible fee structure. At the macroscopic level, this improves capital efficiency while significantly raises the bar for liquddidity providers from microscopic. In this project, we try to find out some Uniswap V3 LP strategies. Statistical Analysis in Portfolio Construction Supervisor: CHEN Kani / MATH Student: ZHANG Kaizheng / QFIN Course: UROP1100, Summer The background of this UROP project is a recent V3 update of the largest DEX, (decentralized exchange), Uniswap. Specifically, Uniswap used to manage liquidity providers’ stored tokens uniformly. In V3, however, it divides liquidity on different price ranges, and liquidity providers are required to specify the range on which they will provide every piece of liquidity. This fundamentally changes the strategies of liquidity providers – previously LPs could simply profit by putting a static pair in the pool (passive strategy). However, V3 would require LPs to predict the volatility of the relative price, making passive strategies unprofitable. Making a prediction of such near-future volatility and optimize other aspects of the strategy is the key for us when designing a practically profitable “proactive” liquidity management strategy. During the project, several PLM strategies were designed, optimized and backtested. 44 School of Science Department of Mathematics Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: WU Ruirui / MATH Course: UROP2100, Fall This report will briefly present theorems proved in the paper by Hamilton in 1986 and paper by Brendle and Schoen in 2008. These two papers studied manifolds with specific curvature conditions and proved them to be diffeomorphic to some sphere space forms. The report will mainly focus on how the proof of main theorems goes in them, and complete some details omitted by authors. Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: CHEN Zijun / MATH-PMA LIU Lingchong / MATH-PMA Course: UROP3100, Fall UROP3100, Fall In this semester, we went through RICHARD S. HAMILTON’s paper FOURMANIFOLDS WITH POSITIVE CURVATURE OPERATOR. And then discuss definition and application of PIC(Pinched Isotropic Curvature) in MANIFOLDS WITH 1/4-PINCHED CURVATURE ARE SPACE FORMS, by SIMON BRENDLE and RICHARD SCHOEN. We spent most of our time discussing and understanding the ODE-PDE Maximum principle and convergence criterion mentioned in Hamilton’s paper, and then applied it on the dimension 3 and dimension 4 cases. In this report, we will try to summarize above discussed contents from the perspective of logic relations between positivity of different curvatures and then. Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: LI Jiangtao / MATH-PMA Course: UROP3100, Spring In Hamilton’s 1982 paper, he introduced the Ricci Flow equation = −2Ric(g) and proved that the psoitive Ricci curvature is preserved by Ricci Fow, with help of a special maximum principle (the maximum principle for tensor). In this expository paper, we study the effect of Ricci flow on another important curvature condition, nonnegative isotropic curvature and follow Nguyen to prove that it is also preserved by Ricci Flow. However, this curvature condition is not as simple as positive Ricci curvature. To handle with this, we need to introduce a modified version of maximum principle, which allows us to reduce the study of PDE on vector bundle over manifold to system of ODEs on fibers of the bundle. This maximum principle was also first introduced by Hamilton in his 1986 paper and then Nguyen and Brendle, with Schoen used it to prove that nonnegative isotropic curvature is preserved by Ricci Flow. Brendle and Schoen went further and finally proved the differentiable sphere theorem, which we will discuss briefly in this paper but will not present the proof due to limited research time. At the end of the paper, by a modified version differentiable sphere theorem due to Petersen and Tao, we deduced a improved differentiable sphere theorem for odd dimension in Corollary5.4. 45 School of Science Department of Mathematics Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: WU Ruirui / MATH-PMA Course: UROP3100, Spring This report will present a generalization of results in “Rotational Symmetry of Asymptotically Conical Mean Curvature Flow Self-Expanders” by Frederick Tsz-Ho Fong and Peter Mcgrath in 2016 concerning the rotational symmetry of asymptotically conical mean curvature flow self-expanders. The technique of proving main theorem in “Rotational Symmetry of Asymptotically Conical Mean Curvature Flow Self-Expanders” will be emphasized and some detailed calculations are very similar as in [2]. [2] Tsz-Kiu Aaron Chow & Ka-Wing Chow & Frederick Tsz-Ho Fong. “Self-Expanders to Inverse Curvature Flows by Homogeneous Functions”. In: arXiv:1701.03995 (2017). Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: LI Ka Ki / SSCI Course: UROP1000, Summer The report is based on the paper “The Heat Equation Shrinking Convex Plane Curves” written by M. Gage and R. S. Hamilton in 1986. In the paper, the curve shortening flow is defined such that the rate of change of point is proportional to curvature. This report is going to explain the details of some of the theorems and proofs in the paper, and to generalize some of the results to the flow that the rate of change of points are not proportional to curvature. Due to the requirement and limitation of content, the report only covers the theorem until the boundedness of k. Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: MA Xiangyu / SSCI Course: UROP1000, Summer In this paper we show that as a smooth convex plane curve is deformed along its normal vector field at a rate proportional to its curvature, the isoperimetric ratio L2/A of the curve approaches 4π as the enclosed area approaches zero. If the one parameter family of curves evolving in this fashion is normalized by homothetic expansion of the plane so that each curve encloses area g, then the resulting family will converge to the unit circle. 46 School of Science Department of Mathematics Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: TAN Weile / SSCI Course: UROP1000, Summer In this paper, we summarize the basic stuff of plane curves and the curve shortening flow, and show that if we deform a convex smooth plane curve at a speed proportional to its curvature along its normal vector field, the isoperimetric ratio L2/A decreases to 4π. If we normalize the family of curves in this transformation by homothetic expansion so that all the lengths of the curves are 2π, then the family we got will converge to the unit circle. Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: WU Haoyu / SSCI Course: UROP1000, Summer This report focus on some basic knowledge about flow geometry. It’s a statement that there is a evolution, i.e. the heat equation, for a closed convex plain curve shrinking to a point with spect to time t ∈ [0,T), where t is finite. The curve is also becoming more and more circular in the evolution. Firstly, this report will state some basic lemma in the calculation of some value of the plane curve. Then mainly, this report gives a way to prove that the curve is becoming more and more circular during the evolution. Using some geometric techniques, we show that the ratio of the maximum curvature and minimum curvature of the curve goes to 1. Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: YEUNG Wai Ho / SSCI Course: UROP1000, Summer This report is about the comparison between Euclidean Geometry and Riemannian Geometry. How does it motivate some geometric idea, for example, define the arc-length and surface area/volume intrinsically. After explain how to generalise the concept of parallel and straight line from Euclidean space to an ambient manifold detailly. For example the intuition of parallel transport and geodesic etc. Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: CHUNG Tsun Ho Anson / MATH Course: UROP1100, Summer Gage and Hamilton first showed that convex curves shrink to a point under the curve shortening flow in 1986. This paper extends such results to a more general case of the curve shortening flow, where the curve’s flow is proportional to the 𝛼th power of its curvature, using the same techniques with suitable modifications. The presented proof is much simpler than Huisken’s two-point estimates or methods involving the support function by Andrews, as it uses only curvature estimates to bound the curvatures. The paper aims to be an introductory essay to the simplest geometric flow without prior knowledge of differential geometry. 47 School of Science Department of Mathematics Geometric Flows Supervisor: FONG Tsz Ho / MATH Student: WU Ruirui / MATH-PMA Course: UROP4100, Summer In this report we will present the proof of the theorem which says finite time existence occurs only if second fundamental form blows up by sorting, paraphrasing and completing the argument in Extrinsic Geometric Flows by Ben Andrews and Bennett Chow and Lecture Notes on Mean Curvature Flow by Mantegazza, together with a summary of chapters about basics of MCF in the former text, in which several propositions are used in the former proof. A Machine Learning Approach to Study the Relationship between Urban Morphology and Urban Heat Island Supervisor: FUNG Jimmy Chi Hung / MATH Student: MOK Wan Hin / SSCI Course: UROP1000, Summer Urban heat island effect is a common phenomenon in cities, with temperature in urban areas being higher than the rural counterparts. It is believed that there is a certain positive correlation between urban morphology and land surface temperature (LST). This project focuses on this phenomenon in the Greater Bay Area (GBA). Over this region, the data of land surface temperature, urban morphology factors, and geographical factors from satellite images were gathered. 2 types of machine learning models – deep neural network (DNN) and linear model - were built, aiming to predict the LST out of these factors. Results show that there is a moderate correlation between urban morphology factors and LST, and that DNN model has a better performance than a linear model in making such predictions. A Machine Learning Approach to Study the Relationship between Urban Morphology and Urban Heat Island Supervisor: FUNG Jimmy Chi Hung / MATH Student: SHREEHARSH Aastha / SSCI Course: UROP1000, Summer In today’s day and age, almost any problem civilization has, or any question humankind wants to answer that deals with a large amount of data can be solved using machine learning (ML) and computer vision. This project is a new attempt to utilize deep learning techniques in order to glean insights on building data and draw conclusions about the Urban Heat Island (UHI) effect based on the urban morphology without simplifying the data or narrowing it down to regional impact. This report focuses on the potential of obtaining this building data using a convolutional neural network (CNN), specifically the U-net Xception style of semantic segmentation. Multiple attempts to improve the accuracy of results produced by modifications to a pre-existing model architecture are outlined alongside recommendations for future work on the same. 48 School of Science Department of Mathematics A Machine Learning Approach to Study the Relationship between Urban Morphology and Urban Heat Island Supervisor: FUNG Jimmy Chi Hung / MATH Student: CHEUNG Cheuk Lun Alan / MATH-IRE Course: UROP1100, Summer Gas chromatography-mass spectrometry (GC-MS) is one of the most commonly used methods to analyse the trace of organic pollutants in the air. In Gas chromatography, sample gas passes through a capillary column that separates different components in the sample gas, resulting in different retention times of different components. Combined with mass spectra, GC-MS shows the various components inside sample gas and their abundance. A Machine Learning Approach to Study the Relationship between Urban Morphology and Urban Heat Island Supervisor: FUNG Jimmy Chi Hung / MATH Student: HE Qihao / DSCT Course: UROP1100, Summer Urban Morphology involves various continuous features such as building height and area that correlate closely with Urban Heat Island (UHI) phenomenon and Land Surface Temperature (LST). This study applies the Feedforward Neural Network (FNN) machine learning approach in an attempt to quantify the relationship between 9 urban variables and LST in the Pearl River Delta region. Various sets of hyperparameters have been tested out. A 9-32-16-1 network structure is proved to outperform the traditional linear regression model by comparing the correlation coefficient between test set actual LST values and predicted values produced by the model. Further improvements in the current NN model are possible with improvements on data and model hyperparameters. Retrieving of Urban Morphology from Satellite Images and Machine Learning Supervisor: FUNG Jimmy Chi Hung / MATH Student: MAK Chi Yuen Kevin / DSCT Course: UROP1000, Summer Hydrochar is a product of the hydrocarbonization of biomass, which is obtained in the form of slurry, a mixture of liquid and solid. It could be produced with either dry or wet biomass, with water added during the hydrothermal carbonization process. This report considers the effects on soil properties when hydrochar with coffee ground as feedstock is mixed into completely decomposed granite soil (CDG), as soil amendment is known to be an important application of hydrochar. Apart from the hydrochar, similar experiments are carried out with its feedstock. This report would compare their effects on soil strength, which appeared to be rather diverse in terms of soil cohesion and friction angle. 49 School of Science Department of Mathematics Retrieving of Urban Morphology from Satellite Images and Machine Learning Supervisor: FUNG Jimmy Chi Hung / MATH Student: YANG Yuang / DSCT Course: UROP1100, Summer Urban morphology is a study that has a valuable application on meteorological modeling, carbon emission estimation, city environment studies, and city planning. Among all urban surface composition information, the information of buildings in the city is especially important. This project aims to develop a machine learning approach to retrieve building locations and dimensions from satellite images, which would significantly reduce the cost of gaining urban building data and facilitate environmental studies worldwide. In the first phase of this project, a model based on FCN-restnet101 is used to handle the semantic image segmentation task to exploit satellite images to generate a bitmap that implies the locations of buildings. After a short session of preliminary training, the model achieves an average overall precision of 80%. Data processing and techniques to prevent model outputting the same value are also discussed. Use Machine Learning Technique to Retrieve Air Quality Data of SO2 Supervisor: FUNG Jimmy Chi Hung / MATH Co-supervisor: LU Xingcheng / ENVR Student: WU Zhiang / COSC Course: UROP1100, Fall Sulphur dioxide (SO2) is one of the most important factors of air quality, which takes part in some chemical reactions in the atmosphere. Prediction based on past observation is on the topic of environmental science. With the development of machine learning in computer science, more and more deep learning techniques are applied in this area. In this project, we tried to use an ensemble BP neural network to predict SO2. Many experiments are conducted to find a reasonable and good model with the data provided by some scientific institutions. Use Machine Learning Technique to Retrieve Air Quality Data of SO2 Supervisor: FUNG Jimmy Chi Hung / MATH Co-supervisor: LU Xingcheng / ENVR Student: YI Sophia / COMP Course: UROP1100, Fall Sulfate and nitrate constitute a significant portion of air pollutants emitted into the atmosphere. Source apportionments of have shown that among many factors including industrial emissions, biomass burning, and residual combustion, secondary sulfate and secondary nitrate were the most abundant sources of PM2.5(Huang et al., 2018), contributing to 33% and 23% of PM2.5 mass annually (Wang et al., 2019). Therefore, monitoring the concentration of sulfate and nitrate in the air is crucial for the management. This project builds up a backpropagation model to estimate and in 2005 and 2015 based on 13 parameters including meteorological variables and pollutants variables. By training these datasets, the model is able to reflect secondary sulphate and secondary nitrate over a particular region. 50 School of Science Department of Mathematics Use Machine Learning Technique to Retrieve Air Quality Data of SO2 Supervisor: FUNG Jimmy Chi Hung / MATH Co-supervisor: LU Xingcheng / ENVR Student: SUN Haochen / DSCT Course: UROP2100, Fall In this report, we will explore multiple deep learning model structures based on the Long Short-term Memory (LSTM) layers that predict the concentration of FSPMC and O3 in the air. With ground observation data of weather and air quality for the past 72 hours as the input to a baseline model, we try to predict the FSPMC and O3 concentration of the future 48 hours. We modify the model by adding WRF-CMAQ prediction of the target hours as an additional input, applying 1D-CNN layers to the input, and applying non-trivial last-layer activation functions and loss functions. With the methods listed above, we achieve a relatively high accuracy in the task. Use Machine Learning Technique to Retrieve Air Quality Data of SO2 Supervisor: FUNG Jimmy Chi Hung / MATH Co-supervisor: LU Xingcheng / ENVR Student: SHA Yu Hin / COSC Course: UROP3100, Fall Air pollution has been a crucial problem for its adverse health and economic impact, hence the need for precise air quality forecasting. With advancements in machine learning, neural networks have been a dominant tool for air quality forecasting due to their high time series prediction accuracy. Models, including recurrent neural networks and Seq-2-Seq networks, have been thoroughly experimented with in the field. A novel network architecture called Transformer has been proposed for performing machine translation. In this project, we adopted this Transformer approach for making 24-to-24 hours forecasting in PM2.5 and O3 forecast and assessed its performance through 53 air monitoring stations across China. Use Machine Learning Technique to Retrieve Air Quality Data of SO2 Supervisor: FUNG Jimmy Chi Hung / MATH Co-supervisor: LU Xingcheng / ENVR Student: YI Sophia / COMP Course: UROP2100, Spring Fine inhalable particles, with diameters that are generally 2.5 micrometers and smaller (𝑃M2.5 ) are the main component of air pollutants emitted to the atmosphere. They are the major cause of reduced visibility. Moreover, exposure to such practices poses a threat to health. Therefore, monitoring the concentrations of 𝑃M2.5 is crucial for air quality management. Obtaining accurate estimates of 𝑃M2.5 from the ground monitoring sensors is a challenging task due to the limitation on the resolution of measured data. So, machine learning is taken to compensate for such disadvantages. The previous approach of using a backpropagation network for prediction achieved relatively good performance, but the performance could be further improved. A transfer learning approach based on the convolutional network is taken in this report. This project builds up the model to estimate PM2.5 from 2015 to 2017 using 15 parameters including Aero-optical Depth. By training these datasets, the model may reflect the concentration of 𝑃M2.5 in China. 51 School of Science Department of Mathematics Modeling Neural Circuit Dynamics in Whole-Brain Imaging Data of Larval Zebrafish Supervisor: HU Yu / MATH Student: CHAU Yu Hei / DSCT ZHANG Ke / MATH-PMA Course: UROP2100, Fall UROP2100, Fall The purpose of this report is to document the analysis we have conducted in the UROP project in the Fall semester (2020-2021). There have been two main areas of focus: the zebrafish prey capture analysis part and the gating model part. Our methods are by no means novel. Most of the analysis are conducted using existing methods, for example, by using available software packages such as MATLAB. Some of them are technical in nature, such as visualizing neuronal information using a 2D image plot. There are also reproduction of results established in other papers, so as to test hypotheses with the given models. The relation between some of the analyses below may be small, as they are minor fragments inside the zebrafish project. This report consists of sections that documents a particular aspect in the project. In every section of this report, the individual goals of the corresponding section will first be stated, and then the details such as methodology and usage will be written. RNN for Biological Neural Networks Supervisor: HU Yu / MATH Student: CHEN Zixin / DSCT Course: UROP1100, Fall Artificial Recurrent Neural Networks such as LSTM and GRU are highly versatile and achieve breakthrough performance in many applications of sequential data. However, the RNN structure for neural science is different from the common one in other fields. Thus, in this project, we customize a RNN for neural science and use it to model the complex temporal dynamics of real neural networks in the brain and explore variations of RNN that are biological plausible with interpretable parameters that can be potentially linked to anatomical structures and biophysical properties. After integrated three Bio-RNN related papers, we built a special RNN and explored its properties on three tasks related to neural science. The results of these experiments prove the efficiency of our RNN. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: CHOY Sin Hang Sonia / MATH-IRE Course: UROP2100, Fall Cluster algebras have been linked to snake graphs, continued fractions and surface triangulations in separate papers. At the heart of cluster algebra is the mutation of a cluster variable, where one can mutate one variable to obtain a new one. This report establishes the correspondences between cluster algebras and positive continued fractions, and the effect of a cluster mutation on the corresponding continued fraction, via their links with triangulations and snake graphs with weighted edges. 52 School of Science Department of Mathematics Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: SAMIN Thanic Nur / MATH-PMA Course: UROP2100, Fall In this report we discuss the connection between Cluster Algebra and Jones Polynomials from knot theory. Cluster Algebra is related to Continued Fractions, in the sense that we can study continued fractions as quotients of cardinalities of perfect matching of certain graphs, that appear naturally in Cluster Algebra. The graphs in question are called Snake Graphs. A class of knots: the two bridge link, can be mapped to continued fractions. The main result is that, in terms of normaliztion upto the leading term, the Jones polynomial of the knot is equal to the F-polynomial of the cluster variable under a specific substitution. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: CHOY Sin Hang Sonia / MATH-IRE Course: UROP3100, Spring Most of this report aims to fill in the gaps of [1], which explains the relationship between cluster algebra mutations and complex volumes of knots. The dilogarithm function provides a complexification of the usual volume function, generalizing to the Bloch-Wigner function. Cluster algebra is used to give a combinatorical interpretation of the R-operator; each cluster mutation corresponds to a tetrahedron. Four tetrahedrons combine to become a hyperbolic octahedron, which are, in turn, associated to knot crossings, and are ultimately used to compute the complex volume of a knot. We expand on Section 4 of [1], which gives the numerical confirmation for Conjecture 2.4 in the two simplest cases, that of the figure-eight knot and trefoil knot. [1] K. Hikami, R. Inoue, Braids, complex volume and cluster algebras. Algebr. Geom. Topol. 15 (2015), 2175-2194. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: MAN Ryuichi / MATH-IRE Course: UROP3100, Spring In this project, we first follow [KN11] to survey the basics in the study of the periodicity of cluster algebras and the associated quantum dilogarithm identities. Inspired by the related papers, we are motivated to investigate the conditions in which a quantum dilogarithm identity is a consequence of the trivial and pentagon relations, with the central focus on the cases in which the initial quiver is a square product AkAn. We shall introduce the relevant tools from the study of maximal green sequences, summarize the results from several recent journal articles and elaborate on their relations to the main conjecture (Conjecture 3.1) that we study. [KN11] R. M. Kashaev, T. Nakanishi, Classical and Quantum Dilogarithm Identities, SIGMA, 7, 2011. arXiv: 1104.4630. 53 School of Science Department of Mathematics Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: SAMIN Thanic Nur / MATH Course: UROP3100, Spring In this report, we discuss Cluster Algebra, a function called the dilogarithm function and how they can be used to calculate complex volumes of hyperbolic knot complements. Cluster Algebra is an algebraic structure equipped with cluster variables, frozen variables and an exchange matrix. We can mutate the variables using rules that are dictated by the exchange matrix. These mutations are related to the dilogarithm functions, specially several identities associated with dilogarithm functions. We can also define the R operator from the mutations, which we can use to compute the complex volume of hyperbolic knot complements using the dilogarithm function. Finally, we showcase an example of calculating the volume of a knot complement. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: HO Sui Kei / COSC Course: UROP1100, Summer We review the definition of cluster algebra of geometric type and cluster algebra with coefficients. Quivers and mutation formulas would be introduced to define the cluster algebra of geometric type. The abstract algebraic structure would be used to define the cluster algebra with coefficients. Laurent phenomenon, which is present in all cluster algebra, would be discussed. We then explore the phenomenon by an example. Finally we would describe cluster algebra from triangulation of unpunctured surfaces. We review the explicit formula of any cluster variable in terms of a given cluster with the aid of snake graph and perfect matchings. A concrete example would be given to demonstrate the findings. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: HUANG Luyi / MATH Course: UROP1100, Summer In this paper, we firstly give a brief motivation and definition of cluster algebra with principle coefficient in section 1. Then in section 2, we are going to examine Laurent Phenomenon. In section 3, perfect matchings and related cluster expansion formulas will be studied. In section 4, an example concerning section 3 will be given. 54 School of Science Department of Mathematics Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: KWOK Kin Ming / SSCI Course: UROP1100, Summer Cluster algebra was first introduced in S. Fomin and A. Zelevinsky. “Cluster algebras I: Foundations”. In: Journal of the American Mathematical Society 15 (2001), pp. 497–529, which was related to different area of Mathematics such as Lie theory and classification of Lie group. This report would first introduce cluster algebra from triangulation of convex polygon and Ptolemy’s theorem, and then touch on concept such as quiver, seed mutation, coefficient field and Laurent phenomenon. Then a result from Gregg Musiker and R. Schiffler. “Cluster expansion formulas and perfect matchings”. In: Journal of Algebraic Combinatorics 32 (2008), pp. 187– 209 on deriving the formula to represent any cluster variables in terms of Laurent polynomial of a any cluster using perfect matching would be discussed with an example. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: ZHANG Ke / MATH-PMA Course: UROP1100, Summer In this article, we summarize the study of cluster algebra in UROP1100 during summer 2021. We start with a summary on the lecture notes including the definition of cluster algebra, its relation to quiver mutation and Laurent Phenomenon. Next, we relate the triangulation of a surface to cluster algebra. We present a method turning a triangulation into a graph and build a map between certain paths in the triangulation and perfect matchings in the graph. Using this method, we obtain a formula for the cluster variable and its coefficients, which we illustrate with an example. Cluster Algebra Supervisor: IP Ivan Chi Ho / MATH Student: ZOU Huaiyang / MATH-PMA Course: UROP1100, Summer This report first reviews the basics of cluster algebra. We recall the basic structure of cluster algebra like cluster variables, exchange matrix, mutation, coefficient and so on. And we also recall an important feature of cluster algebra, the Laurent Phenomenon, as well as Caterpillar Lemma which can prove the phenomenon. Secondly, the report talks about the content of the paper. We associate an unpunctured surface with a cluster algebra with principle coefficients and then use the theorem introduced in the paper to give the expansion formula of some cluster variables which are related to arcs on the surface. At the end, an original example of a decagon is given to illustrate the idea above. 55 School of Science Department of Mathematics Quantum Groups Supervisor: IP Ivan Chi Ho / MATH Student: CHUNG Soobeom / PHYS Course: UROP1100, Spring The quantum groups is associated with certain Hopf algebras which are q-deformation of the enveloping Hopf algebras of semisimple Lie algebras. It is useful in areas of mathematics and physics, such as the quantum inverse scattering and statistical mechanics according to Faddeev (1995) and Polyakov (1984). Here, Hopf algebras’ basic structure and Hopf algebras SLq (2) and Uq(sl(2)) will be introduced, which are related to the classical group SL(2). These are some of the simplest quantum groups and their finite dimensional representations will also be covered. Lastly, the braided bi-algebras, which induce a solution of the Yang-Baxter equation and guarantee its existence, will be introduced as they can be used to obtain quantum groups GLq(2) and SLq(2). Quantum Groups Supervisor: IP Ivan Chi Ho / MATH Student: CHAN Hong Ming / COSC Course: UROP1100, Summer This report will introduce quantum group by constructing it from the definition of algebra, coalgebra, bialgebra and Hopf algebra. In particular, we are interested in the Hopf algebra Uq(sl(2)) and its representation. Most of the material studied are from Kassel’s book. Some more information on representation theory can be found in Humphrey’s book and we have studied through it before going to [2]. Also an example of using quantum Clebsuh-Gordan formula to find the basis of a 20 dimensional representation of Uq(sl(2)) on Vn ⊗ Vm with n = 4 and m = 3 is given. [2] Christian Kassel. Quantum groups. Springer-Verlag New York, Inc, 1995. Quantum Groups Supervisor: IP Ivan Chi Ho / MATH Student: PACHECO RODRIGUEZ Wemp Santiago / MATH Course: UROP2100, Summer In this report we will study a possible direction for the extension of the classical quantum link invariants introduced by Turaev in 1988 to the setting the modular double . For this sake we will study the ”positive representations” of the modular double and the corresponding universal R-operator as defined by the work of Faddeev, Kashaev and Bytsko-Teschner. 56 School of Science Department of Mathematics Integro-Differential Equations: Theory and Applications Supervisor: JIN Tianling / MATH Student: SZETO Chun / SSCI Course: UROP1000, Summer The eigenvalues and eigenfunctions of the fractional Laplacian have been of great interest, for both mathematical and physical reasons. In the following, two fundamental properties of the fractional Laplacian will be proved: 1. the first eigenvalue is simple and positive; 2. the first eigenfunction does not vanish on its domain U, which is a bounded, Lipchitz open set. As the first eigenfunction is continuous, without loss of generality one usually assumes that the first eigenfunction is positive. The proofs mainly build on a well-known compactness result of the fractional Sobolev space. Then, one can mimic the corresponding proofs for the Laplacian (−Δ) to prove similar results in the fractional case. Efficient Algorithms for Visualizing Dynamical Systems Supervisor: LEUNG Shing Yu / MATH Student: ZHOU Siyuan / DSCT Course: UROP1100, Fall Using Lagrangian approaches to calculate a point’s (denoted by P(x0,y0)) finite-time Lyapunov exponent (FTLE), we need to know its terminal condition after time T and the behavior of points near P. The approach requires a velocity field of any point to construct an ODE to figure out the terminal condition. However, in reality, we can only collect discrete random points at different time levels, which means we do not know the explicit expression of the velocity field. Now we want to use the Gaussian basis function to approximate the FTLE. The approach is as follows: 1. Collect a set of points at different time levels. Use (x(T)-x(T0))/dt to approximate each point’ s velocity at different time level. 2. Getting a set of points with velocity in step 2, use the Gaussian basis function to reconstruct a velocity field. 3. Use the reconstructed velocity field to develop the ODE function to get the terminal condition of different initial points. Calculate FTLE of different points. This report will use a known velocity field to get a set of random points at different time levels, then follow the above steps to get an approximated FTLE. Efficient Algorithms for Visualizing Dynamical Systems Supervisor: LEUNG Shing Yu / MATH Student: QIN Hanmo / MATH-CS Course: UROP2100, Spring The backward phase flow method is an Eulerian approach to approximate the Lyapunov Exponent of a dynamic system. The method uses interpolation to evaluate the midway short time flow maps of a given long time flow map. A suitable interpolation method means the map will be closer to reality and is more accurate. In this report, we will discuss and compare several different interpolation skills and conclude which one is more suitable for the moderation of the backward phase flow map. 57 School of Science Department of Mathematics Efficient Algorithms for Visualizing Dynamical Systems Supervisor: LEUNG Shing Yu / MATH Student: CHAU Wai Ming / MATH-SFM Course: UROP1100, Summer We present a principal component analysis (PCA) method for computing the finite time variance analysis (FTVA) in uncertain vector fields. The definition of FTVA is developed from the finite-time Lyapunov exponent (FTLE) for deterministic vector fields. We first do the simulations by solving the corresponding stochastic differential equation. The method allows us to get the maximum eigenvalue from the covariance matrix, which measures the arrival location's maximal variance direction. Moreover, we explored the weighted principal component analysis (WPCA) method based on the diagonalization of the weighted covariance matrix. We will apply the methods to the double gyre flow to demonstrate the performance of the two methods. Efficient Algorithms for Visualizing Dynamical Systems Supervisor: LEUNG Shing Yu / MATH Student: TSAI Yun Chen / MATH-IRE Course: UROP1100, Summer Lagrangian coherent structure (LCS) is important for studying dynamical surfaces, and the finite-time Lyapunov exponent (FTLE) is an essential quantity in analyzing the LCS. The current method of visualizing the FTLE field is inefficient in extracting the ridges as a significant fraction of computation resources have been wasted on unrelated regions. We propose a quadtree-based adaptive refinement method to visualize the FTLE field on a 2D plane and test it with the double gyre flow. Also, we develop a rough error estimation to show the effect of the adaptive refinement. Efficient Algorithms for Visualizing Dynamical Systems Supervisor: LEUNG Shing Yu / MATH Student: YAU Wing Yan / SSCI Course: UROP1100, Summer We investigate how infinitesimal portion (area for 2d flow, volume for 3d flow) changes over time in fluid flows. The project tests the performance of two different Lagrangian approaches. The first one approximates the final positions of grid points to obtain the Jacobian matrix, while the other one keeps track of the deformation of the infinitesimal portion. Both methods give results that can show the details of flow structure, yet the former one would have higher accuracy when more refined grids are applied while the latter one has stable performance. 58 School of Science Department of Mathematics Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: SHUM Ka Shun / DSCT Course: UROP1100, Fall Story Completion is a challenging task of generating the missing plot for an incomplete story. In this report, we give a generative adversarial networks based on Transformer (T-CVAE-GAN) for the missing sentence generation. Our approach uses shared attention layers for Transformer encoder and decoder and a latent variable provided by a generator. Through this generator, we could generate diverse reasonable plots. BLEU eDvaluation is used to show that our model can achieve state-of-the-art results in terms of readability, diversity and coherence. Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: WU Hsuan-cheng / MATH-IRE Course: UROP1100, Fall The agent learns a model of the environment. The advantage of the modelbased is that agent is allowed to plan by thinking and analyzing ahead since it has a model, and then turn into a learned policy. If it works, then the speed of improvement is more efficient than those methods that don’t have a model. On the other hand, the disadvantage of modelbased is that a groundtruth model of the environment is usually not available to the agent, then the agent need to learn the model purely from its experience. It is highly possible that there will be a bias of the model and the environment as time passed, so the optimal behaviors are not the same as what the agent learns. The most famous example of the modelbased methods is AlphaZero. Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: ZHONG Dingyan / DSCT Course: UROP1100, Fall Object detection is one of the heated topics in computer vision nowadays. Plenty of algorithms based on deep learning have arisen over the past few years. Typically, a 2D object detection method uses either anchors or key points to estimate the locations of objects. The difference between those two kinds of methods is only the level of abstraction. 3D object detection yields more difficulties as it concerns depth and rotation angles, which are hard to predict from one image, especially in a monocular case. This project is organized by reviewing and reproducing existing algorithms, together with proposing solutions to the remaining problems. 59 School of Science Department of Mathematics Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: SHUM Ka Shun / DSCT Course: UROP2100, Spring Conventional context-aware text generation only focuses on What to generate when given the surrounding texts. Inspired by existing corrupted long-form narrative, we propose new setting of detection and generation where the model need to find Where to generate and then What to generate. The task is more challenging since it needs to detect where exits possible semantic gap and then infer and generate. Then we introduce our Two-stage prompt-based generation framework and do experiments on ROCStory dataset. The result shows the effectiveness of our model to handle such new task and generate reasonable answers. Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: RUBAB Tamzid Morshed / DSCT Course: UROP1100, Summer This is a progress report describing the contents I have learned over the summer. These include various neural network architectures, training algorithms, implementing them with pytorch, several 2d object detection algorithms such as R-CNN, fast R-CNN, and faster R-CNN. In this report, I will first describe neural networks. This part includes data collection and pre-processing, designing a neural network model, choosing loss functions, back-propagation, training, and convolutional neural network. Then I will describe some 2d object detection algorithms by reviewing the papers of R-CNN, fast R-CNN, and faster R-CNN. Faster R-CNN was the first step towards real-time detection, which is very important in tackling various modern problems. Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: XU Yanbo / COSC Course: UROP1100, Summer Object detection and localization has gain enormous attention in recent years, due to its potential application in autonomous driving, robotics, and so on. Among those many methods, Monocular Three-Dimension Object Detection is a challenging problem because of its lacking of depth information in the input, yet human is capable of estimating distance and locating objects with prior knowledge. With the development of deep learning and developing research on 2D object detection, depth estimation, and so on, 3D detection has become feasible. Meanwhile, 3D annotations in often hard, and other types of input data (such as lidar signals and Multiview images) are expensive to deploy, making the supervised approaches limited by data. In the context of autonomous driving, the most common type of data is monocular video. Therefore, in this research project, we propose a semi-supervised monocular 3D detection framework, to make good use of both annotated data and raw videos. 60 School of Science Department of Mathematics Research in AI and Machine Learning Supervisor: ZHANG Tong / MATH Student: LI Xinyuan / QFIN Course: UROP2100, Summer In the summer UROP project, I learned basic knowledge about natural language processing through courses and common tools, also read & replicated a couple of papers, all within the area of NLP applications in finance, covering both the stock market and the capital market. I realized that with the abundance of data sources: Twitter data, various Financial statements, index data, there’s a wide range of tasks that NLP can perform, just that I need to do some additional work in data-preprocessing. In the end, I proposed a few experiments based on the papers I replicated (mostly experimenting with loss functions and sub-models), further experiments will continuously be conducted in the future. 61 School of Science Department of Ocean Science Department of Ocean Science Phytoplankton Biomass and Community Composition in Hong Kong Coastal Waters Supervisor: LIU Hongbin / OCES Student: LI Hau Ki / OST Course: UROP1000, Summer Microzooplankton is important for the marine ecosystem, and it includes many groups of organisms such as ciliate and dinoflagellates. Different locations in the Bering Sea have different conditions, such as chlorophyll concentration. So, the abundance and biomass of microzooplankton may vary in different locations respectively. To study the abundance and biomass, samples were collected from different stations in the Bering Sea and examined by using a microscope. Moreover, the abundance and biomass of microzooplankton among different locations in the Bering Sea will be compared. This report concluded that the abundance and biomass are the highest in Anadyr Bay, which is at the northwest Bering Sea. Study on the Adaptation and Physiological Responses of Synechococcus Isolates from Different Oceans Supervisor: LIU Hongbin / OCES Student: IO Kai Hong / ENVS LAM Chung Ling / ENVS Course: UROP4100, Fall UROP1100, Fall Synechococcus as the second most abundant oxygenic phototroph in the ocean are known to exploit a wide range of light niches. This arises from the high diversity of pigments among Synechococcus. While different light environments can accommodate strains of various pigment types (PTs) of Synechococcus with different pigment compositions, some strains are found to be capable of Type IV chromatic acclimation (CA4), where the content of their phycobilosomes changes according tot the ambient light quality. Different strains of Synechococcus growing in white and blue light were studied to investigate the process of chromatic adaptation in Synchococcus. It is found that the pigment composition of Synechoccous changes with the light quality, and some features of possibly strains that can perform CA4 are spotted. The study provides insights of one of the important adaptations of Synechococcus, light acclimation, and studying on the process and characteristics of light adaptation helps understand the distribution and ecology of Synechococcus. Ising Model as a Model for Ice Melt Ponds Supervisor: MAK Julian / OCES Student: LEUNG Hoi Yan Anastasia / PHYS Course: UROP1000, Summer Efforts have been made to understand the formation of ice ponds in the Arctic Sea, to attempt to slow down the melting of ice. There have been postulations that the ice pond formation has similarities to the Ising model. This project simulates the Ising model by Monte Carlo simulation, using MetropolisHastings algorithm and Glauber Dynamics. The change of energy, magnetisation, specific heat, and magnetic susceptibility, with temperature at different magnitude of external field is also computed. The critical temperature of magnets is seen to increase with increasing external magnetic field. This study also attempts to recreate ice pond formations using the Ising model. 62 School of Science Department of Ocean Science Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: ABBAS Rauszen Ocean Annika / SSCI Course: UROP1100, Fall This study has investigated how benthic composition can affect oxygen concentration and net ecosystem production at 3 sites in Palau. Ranging from sand to coral and algae dominated ecosystems, the gradient flux approach was adopted to record oxygen from two heights above the seafloor for 24 hours. A relationship was found that showed that changes in coral and algae coverage resulted in different oxygen production and respiration. The findings reveal how benthic autotrophs can influence the net ecosystem production of coral reefs and why it is vital to better understand how organisms play a part in primary production and how we may predict reef trends in the future. Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: CHU Yui Kei / BCB Course: UROP1100, Fall Coral could be easily damaged by anthropogenic activities and bleaching events. Reduction in percentage coverage of coral over the benthic surface might affect primary productivity and respiration in coral reef. Net ecosystem production (NEP) were measured at five sites in Palau: A1, A4, B1, B4 and L3 through the gradient flux approach. A1 and A4 are coral-dominated reef; B1 and B4 were algal reef and L3 was a sand-dominated reefs. The result points to a possible reduction in respiration rate over coral community as coral reefs transform into algal reefs. The respiration rate of reef during night-time might become more stable. However, the impact over the primary productivity over the reef could not be determined. Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: CHUNG Shun Lap Benedict / ENVS Course: UROP1100, Fall There have been ongoing efforts in coral reef research aiming to understand and protect these visually appealing ecological sanctuaries, under the ever-elevating pressures from innumerable artificial impacts on the marine environment. Sensor monitoring methods for measuring reef metabolic data are often a useful non-invasive, and continuous tool in deciphering the health condition of reef ecosystems. So, this report aims to deliver a trial observation of the net ecosystem productivity (NEP) over several reef sites surrounding Palau. The NEP computation made use of the technique of measuring the oxygen gradient flux released from the primary productivity of the boundary phase of the benthic reef community, in a vertical profile, adapted from Takeshita et al., 2016. Finally, we will discuss the observations made there and its possible implication with reef systems and its species compositions. 63 School of Science Department of Ocean Science Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: LEUNG Shu Kiu / ENVS Course: UROP1100, Fall Coral reefs, endangered by human activity and climate change, are vulnerable ecosystems that demand our attention. To better our understanding of the threats faced by the ocean, it is vital to develop new methods for monitoring reef health. The gradient flux (GF) boundary layer approach in this paper helps to measure a key parameter, net ecosystem production (NEP), for determining ecosystem functions of 5 sites in the reefs of Palau. Within the study area, site A4 (13.33 ± 17.27 𝑚𝑚ol 𝑚−2 ℎ𝑟−1 ) had the highest average NEP, suggesting that the reef harbored the largest number of photosynthetic organisms, and likely algal dominated. Future studies should include net community calcification (NEC) as well to ensure an accurate prediction of reef health. Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: NANDI Shrinivas / ENVS Course: UROP1100, Fall Coral reefs are facing a grave future as they are exposed to severe environmental stresses due to increasing anthropogenic activities. Measurements of Primary Production over coral reefs provide a strong indication of reef condition. Consistent and long-term measurement of these factors over a period of time can indicate the damage being caused by phenomena like Climate Change and Ocean Acidification. In this study, we use the Gradient Flux method in the biodiversity-rich ecosystem of Palau to measure the Primary Productivity and the benthic composition. The measurements were made in 5 different sites, over a 24 hour period. The differences between Primary Productivity and benthic composition at each of the sites were analysed. Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: GOSWAMI Arnab / CEEV Course: UROP1100, Spring Corals consist of small animals called polyps, which exist in large colonies (reefs). This paper studies the coral species characteristics and the abundance in Hong Kong. A further examination of coral structure formation from calcium carbonate and sediments is conducted and the dissolution of these structures through chemical, physical, and biological methods is discussed. A final discussion is presented to show the effect of human activities on coral communities and the importance of these species to the health of the oceans. 64 School of Science Department of Ocean Science Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: NANDI Shrinivas / ENVS Course: UROP2100, Spring UROP1000, Summer Coral reefs are facing a grave future as they are exposed to severe environmental stresses due to increasing anthropogenic activities. The rising ocean temperatures every summer has resulted in more frequent and deadly bleaching episodes around the globe. It is a necessity for scientists to utilize regular monitoring techniques to identify early signs of coral bleaching. In this report, we utilize coral colour reference cards, SCUBA, and Symbiodinium cell counting to establish a baseline for bleaching of corals at Sharp Island in Hong Kong. Characterising Coral Reef Function across Anthropogenic Gradients Supervisor: WYATT Alex / OCES Student: CHEUNG Chun Fei / SSCI Course: UROP1100, Summer Coral reefs support many marine organisms and provide ecological services. Through stable isotope analysis of different samples (including sediments, deposit feeders like sea cucumber, filter feeders like bivalves, etc.) from different field sites, the marine food web of Hong Kong can be constructed, and anthropogenic impacts across the East-West gradient could be investigated. Stable isotope analysis can be used to reveal the diet, trophic level and geographic location of samples. However, there are both advantages and limitations. Below the procedures carried out in the project will be discussed. Moreover, the basic principle of stable isotope analysis as a tool in ecological research will be included. 65 School of Science Department of Physics Department of Physics Fourier Optics Supervisor: ALTMAN Michael Scott / PHYS Student: LO Tsz Yu / PHYS Course: UROP1100, Fall The aim of the project is to study the relationship between the feature of the image and the object under different setting of the Low Energy Electron Microscope (LEEM). This is meaningful as knowing those relationship can help us to have better observation when using the LEEM. In this project, the Fourier Optics approach is used for the calculation of image formation. The image of an amplitude object, step phase objects and sinusoidal phase objects are calculated using different setting of the LEEM, such as aperture diameter and defocus. The image resolution is first evaluated as a function of the contrast aperture for a phase object. Those calculation showed that there is an optimum aperture such that the resolution is minimized. The contrast produced by the sinusoidal phase object also depends on the aperture size, but is independent of aperture size if it exceeds the optimum size. The result also showed that the having a little defocus can also help better observation. Fourier Optics Supervisor: ALTMAN Michael Scott / PHYS Student: LO Tsz Yu / PHYS Course: UROP2100, Spring A program for simulating images of two-dimensional objects in low energy electron microscopy was implemented. The comparison of images of amplitude and phase objects simulated by this program to results published in reference indicated that the simulation program is working properly. We then used the program to simulate circular and square objects with sizes below and slightly above the instrumental resolution. We found that intensity line scans through the objects in the simulated images may have one or two intensity peaks depending upon their size and the defocus value. These observations may be used to determine the sizes and possibly the shapes of objects below the instrumental resolution. Fourier Optics Supervisor: ALTMAN Michael Scott / PHYS Student: NGUYEN Xuan Tan / SSCI Course: UROP1000, Summer The current model of Fourier optics used in low energy electron microscopy (LEEM) was successful in predicting image formation with high accuracy. However, there are certain assumptions that limit its scope of application to electron sources of high temporal and spatial coherence with Gaussian energy distribution. This report provides a generalisation to spatially and temporally more extended sources by taking into account second-order Taylor expansion of the phase shift caused by source extension. The energy distribution is generalised by a combination of N Gaussian sources, which can be utilised to fit a generic energy distribution. A comparison is made to examine the validity of the current Fourier optics model for the Fowler-Nordheim energy distribution. 66 School of Science Department of Physics The Geometry and Physics of Wrinkling Supervisor: ALTMAN Michael Scott / PHYS Student: CUI Zhenhao / PHYS-IRE Course: UROP2100, Fall As the surface potential of sample can affect the imaging of low energy electron microscopy (LEEM), we can use it to do the surface potentiometry analysis. Here we use it to investigate the surface potential of graphene on electrode made with copper (Cu), silicon nitride (SiN) and free-standing graphene. We report that the voltage drops faster at free-standing graphene when compare to the graphene on SiN. Besides, the current tends to start getting into the graphene near the boundary of SiN and Cu, but not the boundary of graphene. And the surface potential shows a jump which increases linearly with the applied transverse current at that boundary. Evidence for Isotropic s-Wave Superconductivity in High-Entropy Alloys Supervisor: JAECK Berthold / PHYS Student: LEUNG Ka Wun Casey / PHYS-IRE Course: UROP1000, Summer High-entropy alloys (HEA) form through the random arrangement of five or more chemical elements on a crystalline lattice. Despite the significant amount of resulting compositional disorder, a subset of HEAs enters a superconducting state below critical temperatures Tc<10 K. The superconducting properties of the known HEAs seem to suffice a Bardeen-Cooper-Schrieffer (BCS) description, but little is known about their superconducting order parameter and the microscopic role of disorder. We report on magnetic susceptibility measurements on films of the superconducting HEA (TaNb)1−x(ZrHfTi)x for characterizing the lower and upper critical fields Hc1(T) and Hc2(T), respectively as a function of temperature T. Our resulting analysis of the Ginzburg-Landau coherence length and penetration depth demonstrates that HEAs of this type are single-band isotropic s-wave superconductors in the dirty limit. Despite a significant difference in the elemental composition between the x=0.35 and x=0.71 films, we find that the observed Tc variations cannot be explained by disorder effects [1]. [1] Casey K.W. Leung, Xiaofu Zhang, Fabian von Rohr, Rolf Lortz, and Berthold Jäck, arXiv:2111.11013 [condmat.supr-con] (2021). Machine Learning Aided Detection of the Quantum State in an Atomic Quantum Simulator Supervisor: JO Gyu Boong / PHYS Student: MAK Ting Hin / PHYS-IRE Course: UROP3100, Fall A convolutional neural network (CNN) is applied to identify the topological quantum phase transition in a system of 173Yb atoms trapped in an one-dimensional optical Raman lattice, based on the insights provided in the previous reports. The CNN is able to identify both experimental transition points by plotting its predictions on a test dataset. Visualizing the filters of the convolution layers suggests that the trained CNN classifies by letting each filter Fk convert the datapoint to a single quantity Gk, which reflects the topological state of the system. The results suggests that it is possible to understand how a trained CNN works by properly analyzing the model. 67 School of Science Department of Physics Machine Learning Aided Detection of the Quantum State in an Atomic Quantum Simulator Supervisor: JO Gyu Boong / PHYS Student: MAK Ting Hin / PHYS-IRE Course: UROP4100, Spring Classifying topological quantum phases have generically been a key challenge in quantum many-body physics due to the absence of local order parameter. Here, we report a successful classification of experimental images with different symmetry-protected phases by training a deep convolutional neural network (CNN). The whole phase diagram mapped out with the trained CNN gives consistent phase transition points with previous results extracted by the conventional method. Visualizing the filters and postconvolutional results shows that spin-imbalance information is important for the neural network to make the classification. Our work points out the potential of using machine learning techniques to physically analyze complex systems and large experimental datasets. Electromagnetic and Acoustic Metasurfaces Supervisor: LI Jensen Tsan Hang / PHYS Student: YANG Sandra Sang Xiao / PHYS Course: UROP1100, Fall Optical properties of metal nanomaterials can be tuned using different metals and structures. The conduction electrons of the metal can be excited under visible and near-IR light. When the electrons are collectively excited, they form coherent oscillation. This is known as the localized surface plasmon resonance. The frequency of the resonances varies with its shape, dimensions, and the surrounding dielectric background. Since the resonance frequencies are very sensitive to the dielectric environment and are easy to detect, it can be used for sensing in the nanoscale. This report will discuss the plasmon resonant frequencies of a thin film of modified Sierpinski triangle. Electromagnetic and Acoustic Metasurfaces Supervisor: LI Jensen Tsan Hang / PHYS Student: HO Tin Seng Manfred / PHYS-IRE Course: UROP1100, Summer Solving the reverse scattering problem involving elastic waves means to be able to tell the density distribution and geometry of a material from its scattered waves. This report demonstrates an experimental approach where the scattering of elastic waves on a sample metamaterial can be accurately mapped and analyzed, subject to only a few experimental errors. This can serve as a benchmark for other methods of analysis of the problem, drawing comparisons between theoretical algorithms, machine learning techniques and experimental data. 68 School of Science Department of Physics Time-Varying Metamaterials Supervisor: LI Jensen Tsan Hang / PHYS Student: YIP Heung Kit / PHYS-IRE FAN Wai Hin / PHYS-IRE Course: UROP1100, Summer UROP2100, Summer This report aims to experimentally study the properties of a digitally virtualized acoustic metamaterial first proposed by Wen et. al.. We attempt to reproduce Wen et. al.’s temporal effective medium on an expanded analogous experimental platform with more metaatoms than in the original paper, and attempt to empirically verify our own predictions that a bandgap exists in this metamaterial. We produce and discuss evidence that indicates the realization of a temporal effective medium on our platform similar to Wen et. al., as well as evidence that suggests that the predicted bandgap exists, provided that the metamaterial contains enough metaatoms to enter the effective medium regime. Monte Carlo Simulation of 2D Supramolecular Assembly Supervisor: LIN Nian / PHYS Student: HUI Lik Hang Alpha / SSCI Course: UROP1000, Summer This is a report on simulation with molecular assembly using Kinetic Monte Carlo simulation with Fortran being the language used for the core of the program and Matlab, Excel or other program or platform as support on data analysis. We would be mainly talking about some basic concepts including the equations involved in the simulation of atoms on plane and basic assumptions, how some functions are coded with the logic behind and also some final simulation results with the program developed and conclusion of these data. Quantum State Calculation of Two-dimensional Supramolecular Nanostructures Supervisor: LIN Nian / PHYS Student: CHAN Pak Yin / PHYS Course: UROP1100, Fall This project is aimed to simulate a system of a two-dimensional supramolecular nanomolecule using the MATLAB program. However, there are several unexpected errors due to the MATLAB program itself. To put it briefly, the output parameters (pillar size and pillar potential) after Fourier Transformation has a certain variation from the input values which significantly affect the outcomes. In this paper, the error affecting the pillar potential would be discuss only, while the quantitative analysis for the errors as well as the solution would be conducted. Meanwhile, it is found that the radius difference over layer-to-layer will cause the extra potential difference over layer-tolayer. The qualitative adjustment will be discussed since the construction of the flat band due to an individual change of radius of each layer would also been discussed. 69 School of Science Department of Physics Quantum State Calculation of Two-dimensional Supramolecular Nanostructures Supervisor: LIN Nian / PHYS Student: CHAN Wai Ming / PHYS Course: UROP1100, Fall In this paper, a MATLAB program is used to simulate the physical system of a 2-dimensional supramolecular nanomolecule, by numerically solving the 2-dimensional Schrödinger equation. A four-layer hexagonal lattice cell was constructed in MATLAB. From the program, variables such as band structure, total density of states (TDOS) and local density of states (LDOS) can be measured. Then we would observe the variables’ response to the change of parameters, in this paper, mainly positions of potential pillars. Quantum State Calculation of Two-dimensional Supramolecular Nanostructures Supervisor: LIN Nian / PHYS Student: YUE Cheuk Kan Kelvin / PHYS-IRE Course: UROP1100, Summer In this paper, a physical system of a two-dimensional supramolecular nanostructure is simulated by a MATLAB program, which numerically solves a Schrödinger equation. Hexagonal and triangular potential pillars are constructed and put into the hexagonal lattice in the program. From the program, variables such as band structure, total density of states (TDOS), local density of states (LDOS) and effective mass of electron are calculated. These variables’ response to the change of parameters, including the size r and angle of rotation θ of the potential pillars are then observed. The ultimate goal of this project is to explore how the physical properties of supramolecules are affected by transformation of the structure. Application of Machine Learning in Physics Supervisor: LIU Junwei / PHYS Student: TRAN Duc Huy / PHYS-IRE Course: UROP1100, Fall The Ising model is a simple model which explains the phase transition of ferromagnetic materials. Although the analytical solutions for some simple cases of Ising model has been discovered, investigating more complicated cases still requires numerical simulations. The Metropolis-Hastings algorithm is a popular approach to this problem. Among the different implementations of this algorithm, the local update algorithm is the simplest one and could be generalized to many complex cases of Ising model. However, it faces the problem of auto-correlation. The Wolff-cluster update algorithm could reduce the auto-correlation, but is only valid for the case with two-spin interaction. The Self-Learning Monte Carlo method (SLMC), invented by Junwei Liu, Yang Qi et. al., could combine the advantage of the previous two methods and produce more efficient simulations. 70 School of Science Department of Physics Application of Machine Learning in Physics Supervisor: LIU Junwei / PHYS Student: CHEN You-chiuan / PHYS-IRE Course: UROP4100, Fall Utilization of auxiliary fields is a standard technique in applications of quantum Monte Carlo algorithms when decomposition of many-body interactions between indistinguishable particles is feasible. In this report, we first review how auxiliary fields are introduced with the Hubbard model as example. Next, we discuss how auxiliary fields come with the notorious sign problem, and some conditions under which the absence of the sign problem is guaranteed. At last, we raise a new perspective: even when there is no sign problem, a quantum Monte Carlo algorithm using Metropolis sampling may be less efficient than that using random sampling because of the singularities in Green’s functions induced by auxiliary fields. Theoretical Study of Properties of Real Materials Supervisor: LIU Junwei / PHYS Student: KWOK Yuk Lam / PHYS-IRE Course: UROP1100, Summer We aim at studying about the band structure of graphene ribbon using Python. First, the Hamiltonian from tight-binding model is used, the Bloch Hamiltonian of a 2D graphene is constructed, and its band structure is calculated. Then we do the same thing on graphene ribbon, which band structure depends on its edge shape. We also investigate the effect of ribbon width on the bandgap and edge states. After that, we take the spinorbit coupling into consideration by adding a second term into the Hamiltonian. The strength of spin-orbit coupling is varied, and its effect on the band structures is discussed. Axion-like Dark Matter Supervisor: LIU Tao / PHYS Student: LOU Xuzixiang / PHYS-IRE Course: UROP1100, Spring This report reviews my UROP 1100 project, Detecting Axion-like Dark Matter with Pulsar Polarization Array, in Spring Semester, 2021. As a popular dark matter candidate, axions or axion-like particles are currently under extensive research. Axion-like dark matter in our galaxy behaves like a classical scalar field, and light traveling through it receives a rotation in the linear polarization angle. Based on this effect, we propose to detect axionlike dark matter using linearly-polarized pulsar light, taking into account both temporal and spacial correlations of the polarization angles. We use PPTA pulsar polarization data to set 95% C.L. limits on the axion-photon coupling. 71 School of Science Department of Physics Basics on Dark Matter Supervisor: LIU Tao / PHYS Student: LI Weiran / PHYS Course: UROP1100, Summer For centuries, dark matter has been one of the most attractive topics among physicists, because it seems to be essential for the explanation of various current cosmological observations. In this report, I first introduce some basic concepts that are essential for the understanding of dark matter. Then some important observational facts are briefly summarized and alternative approaches in order to rule out dark matter are introduced. Detailed discussions on the nature of dark matter with observational evidence and researching methods are followed where basics concepts are revised and extended. Finally, a concise introduction of different categories of dark matter candidates is provided. Basics on Dark Matter Supervisor: LIU Tao / PHYS Student: ZHU Jundong / SSCI Course: UROP1100, Summer This report first reviews the basics of cluster algebra. We recall the basic structure of cluster algebra like cluster variables, exchange matrix, mutation, coefficient and so on. And we also recall an important feature of cluster algebra, the Laurent Phenomenon, as well as Caterpillar Lemma which can prove the phenomenon. Secondly, the report talks about the content of the paper. We associate an unpunctured surface with a cluster algebra with principle coefficients and then use the theorem introduced in the paper to give the expansion formula of some cluster variables which are related to arcs on the surface. At the end, an original example of a decagon is given to illustrate the idea above. Setup of an Automated Telescope System as Sensor Testbench Supervisor: SMOOT George F. / PHYS Student: CHAN Yan Yan / PHYS-IRE Course: UROP2100, Fall This report presents the preliminary design of a focal plane cooling system for the automated telescope testbench. The cooling system aims to reduce the dark count noise of a single photon imaging detector which is based on positionsensitive silicon photomultiplier (PS-SiPM). To maintain the photodetector at a low temperature, it is attached to a thermoelectric peltier cooler module, which is further attached to a CPU cooler. A 3D-printed plastic enclosure is also designed to reduce heat gain and disturbance from the external environment. Preliminary testing of the cooling system will be done afterwards. 72 School of Science Department of Physics Stability Analysis in General Relativity Supervisor: WANG Yi / PHYS Student: BIAN Siyu / PHYS WANG Zun / PHYS-IRE Course: UROP1100, Spring UROP1100, Spring We investigate the possible obstacles of ”warp drive” superluminal travels in reality. Instead of discussing the reliabillity of the warp drive spacetime, we consider two factors which may interact with the spacecraft during the interstellar travel. The collision of interstellar dust particles and photons with the spacecraft, as well as the force exerted on spacecraft from the curved spacetime, will impose hylology constraints on the spacecraft. In our paper, we investigate the above two effects in detailed, and give analytical solutions in the limit where the speed of warp drive bubble is much larger than the speed of light. Then we also calculate the tidal force related to two points in the curved spacetime affecting on our spacecraft. Charge Generation and Carrier Diffusion in Organic Solar Cells Supervisor: WONG Kam Sing / PHYS Student: LIU Yidai / SSCI Course: UROP1000, Summer Organic solar cells are generally showing greater voltage loss and thus lower efficiency than their inorganic counterparts. The voltage loss mainly comes from the large energy offset serving as a driving force for charge separation and non-radiative recombination. In the once-popular fullerene-based OSCs, a large driving force was believed to be necessary for efficient charge separation. However, in recent years, some NFA-based organic solar cells which exhibit high charge generation yields with low driving force at the donor-acceptor interface have been developed, reducing the voltage loss. The mechanism behind these systems is still not quite clear. In this review, we try to summarize the possible mechanism of the efficient charge generation in low-energy-offset systems by presenting some of the now-existing theories: (1) the intermolecular electrostatic field is the new driving force; (2) the charge separation is thermally activated; (3) electron wavefunction delocalization at D/A interface reduces Coulomb attraction between electron-hole pairs. Photoexcited Charge Generation and Recombination Dynamics in Perovskite Solar Cells Supervisor: WONG Kam Sing / PHYS Student: KURNIAWAN Michael Alexander / SENG Course: UROP1000, Summer With the reported high stability of Dion Jacobson (DJ) phase 2D perovskites, bringing to light its optical characteristics and charge recombination properties will aid the development for their usage in optoelectronics such as LEDs, solar cells, and laser diodes. This work attempts to find trap densities and characterize the recombination dynamics of two DJ phase 2D perovskites with varying ligand length based on 1, 10–decanediammonium dibromide (DDA) by using time resolved photoluminescence. Photoluminescence quantum yield (PLQY) were carried out to show the much-improved light emission property via the modification of the ligand length. The non-radiative recombination rates were estimated from the experiments, showing that the increase in the rate of non-radiative recombination results in a decrease in light emission. 73 School of Science Department of Physics Neural Information Processing for Highway Latency Prediction Supervisor: WONG Michael Kwok Yee / PHYS Student: HO Man Fung / PHYS-IRE SO Lap / PHYS-IRE JHA Shashwat / PHYS Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall Using data extracted from the Taiwan Highway System, latency of highway journeys between different sources and destination points can be predicted. Significant factors causing the congestion were identified and machine learning models were trained to predict the latency. Previously, a classification method was introduced to preprocess the data into different clusters before training the machine learning model. Subsequently, it was found that by introducing the third dimension into the classification method, the accuracy in the latency prediction was improved. In this report, the effects of different choices of the third dimension on the latency prediction task were studied. Neural Information Processing for Highway Latency Prediction Supervisor: WONG Michael Kwok Yee / PHYS Student: HO Man Fung / PHYS-IRE Course: UROP2100, Spring From results collected from the research last year, it was found that the travel time (latency) between two detectors in Taiwan Highway System can be predicted via an analysis of traffic parameters collected by different detectors using machine learning models, including Extreme Gradient Boosting (XGB) and Gaussian Mixture Model. In this semester, the focus of the research is shifted to ways to increase the accuracy of the prediction by: a revision of inputs and outputs, modification of the training and testing sets and a refinement of the choices of the inputs to the Gaussian Mixture Model. Neural Information Processing for Highway Latency Prediction Supervisor: WONG Michael Kwok Yee / PHYS Student: SO Lap / PHYS-IRE Course: UROP2100, Spring In the task of latency prediction of highway journeys, we examined the error due to the deletion of inputs and studied the mutual information between inputs and outputs. We showed that the accumulation of vehicles in a highway segment and the past latency traversing the segment (and the speed of traversing) have high correlation with the current speed and latency. Compared with other choice of outputs, the mean latency is shown to be having the least mutual information with the inputs among mean, median speed, and latency. In addition, for the inputs to have significant mutual information with the outputs, it requires the segment length per sampling time to be lower than a threshold, which is likely to be related with the speed of vehicles traversing the highway segment. 74 School of Science Department of Physics Experimental Study of The Fast Dynamics in Soft Matter Physics Supervisor: XU Qin / PHYS Student: PENG Huaiyue / PHYS Course: UROP1100, Summer Soft materials often exhibit strong mechanical response to external perturbation in a short period. Many dynamic processes, including droplets impact, pinch-off, and coalescence involve complex mechanics in a time period of a few milliseconds. When a solvent enters the network of crosslinked polymer chains in the gel by osmosis, the gel would experience volumetric growth. Due to the constraint of containers, it will lead to creases and wrinkles on the surface of soft materials. We try to explore the basic nature of instabilities, including their length and distribution. In this program, we will study the characteristics of surface instabilities and use fast camera to study the droplets impact on these instabilities. Interfacial Instabilities of Soft Material Supervisor: XU Qin / PHYS Student: GURUNG Nishan / MATH-GM Course: UROP1100, Summer Swelling Polydimethysiloxane (PDMS) gels can affect their surface properties substantially, which can be of significant research interest given the vast applicability of PDMS. Swelling under surface constraints gives rise to surface instabilities thus, forming micro-patterns. These surface patterns are not well- understood so in this project, we vary parameters such as solvent swelling power, swelling duration, sample stiffness and thickness, and investigate their effects on the observed patterns, including their peak heights. We also report a surprisingly large mass extraction rate via swelling, with the extracted material having a viscosity 6x higher than that of the uncross-linked polymer. This indicates that the swelling process not only extracts uncross-linked polymers, like we normally expect, but also some shorter cross-linked polymer chains. Excitation and imaging of Terahertz phonon-polaritons Supervisor: ZHANG Jingdi / PHYS Student: CHENG Chun Hei / SSCI Course: UROP1000, Summer The goal of this project is to study the effect of a sample placed on a crystal on the propagation of phonon-polariton in the crystal using FDTD (finite difference time domain) simulation which is helpful for the development of phononpolariton spectroscopy. In this UROP 1000 programme, the dispersion relations of LiNbO3 in ordinary and extraordinary polarizations are obtained from the 2D FDTD simulation of phonon-polaritons in the crystal. It is found that using excitation pulses with duration of 1020fs can give the best result for dispersion relation in LiNbO3. The results in this report narrows down the pulse duration range for future development of the simulation and allows comparison with theoretical results to examine the accuracy of simulation. 75 School of Science Department of Physics Femtosecond Laser Pulse Induced Air-Plasma Filament for Generating Thz Radiation: Experiment and Simulation Supervisor: ZHANG Jingdi / PHYS Student: LIN Yuchen / COSC Course: UROP1100, Fall Terahertz electromagnetic wave refers to electromagnetic radiation with the frequency ranging from 0.1~30 THz. Cutting-edge THz technologies, involving generation, modulation, and detection of THz frequency electromagnetic waves are considered to be one of the essential next-generation technologies and the strategic direction of research in the future. Femtosecond Laser Pulse Induced Air-Plasma Filament for Generating Thz Radiation: Experiment and Simulation Supervisor: ZHANG Jingdi / PHYS Student: LIN Yuchen / COSC Course: UROP2100, Spring Following the previous UROP 1100 research, a series of numerical simulations on the amplitude and phase modulation of high-frequency terahertz emission have been performed. Meanwhile, valid models recovering the microscopic dynamics of the amplitude and phase modulation of the terahertz radiation based on the investigation into a kind of phase anomaly have been developed. Also, in preparation for the publication of the research proceedings, illustration figures are also plotted during the project period. In the coming summer semester, we will conduct experiments to demonstrate and verify the validity of our model explaining the mechanism of the generation and modulation of terahertz radiation. Measuring Femetosecond Laser Pulse Duration with FROG Supervisor: ZHANG Jingdi / PHYS Student: YANG Yuang / DSCT Course: UROP1100, Fall Frequency-Resolved Optical Gating (FROG) is the standard method to achieve a complete characterization of ultrafast laser pulses, and the wide adoption of this technique provides convenience and accuracy to labs worldwide. After collecting data points, a retrieval algorithm must be applied to the FROG trace to retrieve the pulse intensity and phase with respect to time. This report focuses on two powerful FROG algorithms, the basic FROG algorithm and the generalized projections (G.P.), reviews some practical considerations for their application and the math and physics behind them, and showcases the pulse characterization from experiment data and data generated from theory using the FROG software by Femtosoft Technology. 76 School of Science Department of Physics Simulation Study of Active Matter Supervisor: ZHANG Rui / PHYS Student: KANG Zheng Tien / PHYS-IRE Course: UROP1100, Summer Active matter comprises large number of active particles, each consuming energy to propagate. Energy is continuously dissipated throughout the movement of individual particles. Various natural phenomena can be modelled via active matter simulations and most of them are biological in origin. Movements of schools of fish, flocks of birds, and bacteria population can be treated as active matters and modelled via active Brownian motion or the Vicsek model. This report seeks to verify the validity of the mean square displacement result produced from the simulation and to measure the radii distribution of micelle concentration emitted from a hypothetical bacterial modelled as an active matter. 77

Abstracts of UROP Projects 2020-21 School of Engineering School of Engineering Department of Chemical and Biological Engineering Department of Chemical and Biological Engineering Development of Functional Food Supervisor: CHAU Ying / CBE Co-supervisor: LIU Yuanshuai / CBE Student: YORDANOV Matey Svetlan / CENG Course: UROP1100, Spring The current paper briefly summarizes the outcomes of the previously conducted USEL and UROP1000 researches on the same topic “Development of Functional Food” and has a goal of translating these outcomes to a successful business plan. For this reason, market and economics considerations were taken into account and a manufacturing process was designed. Moreover, further optimization of this process was considered in order to account for production losses, caused by chipped-out product particles during transportation between the manufacturing line units. As a result, although these losses turned out to be insignificantly small, two additional final products utilizing these particles and involving freeze-dried whelk meat were developed. Consequently, the product range of the company was significantly increased. Development of Functional Food Supervisor: CHAU Ying / CBE Co-supervisor: LIU Yuanshuai / CBE Student: CHEN Jialuo / SENG Course: UROP1000, Summer The UROP course has started for almost 2 months already and I have learnt a lot just through all the research processes. The topic of the project that I am working on is related to beer waste. We want to turn these wastes into useful stuffs such as cat litters and mushrooms. Since beer wastes is purely going to land fill dump without any further reuse which can cause a lot of problems. So, we chose this topic and foresaw the market potential of reusing these wastes as a by-product. During the research process, Professor Liu was able to give me many new ideas and directions which I should be focusing on. Therefore, assisting me to face all the difficulties Nucleic Acid Delivery by Self-Assembled Nanocarrier Supervisor: CHAU Ying / CBE Student: CHANG Cheuk Yan Atta / BIEN Course: UROP1100, Spring Gene delivery is a technique to introduce genetic materials into cells. It is seen as a potential treatment method and its effectiveness is still being explored. These genetic materials are subjected to degradation and delivery systems are needed to protect them from degradation, and to enter a person’s cells for these agents to exert therapeutic effects. Applications of viral-based delivery systems are limited due to immunogenicity concerns, and non-viral based systems are gaining attention. This study aims to explore the transfection efficiency, the stability, and the characterization of peptide-gene delivery system based on a previously discovered short peptide, H10 (Fmoc-PKKKRKVG-HHHH-C12-CStBuHHCStBuCStBuHHCStBuGSPHHD-amide). 80 School of Engineering Department of Chemical and Biological Engineering Nucleic Acid Delivery by Self-Assembled Nanocarrier Supervisor: CHAU Ying / CBE Student: CHANG Cheuk Yan Atta / BIEN Course: UROP2100, Summer mRNA delivery has been a hot research topic. It is seen as a potential treatment method and its effectiveness is still being explored. As mRNA are prone to degradation and are hard to enter and transfect cells on their own, delivery systems are needed to protect them and assist their entrance to a person’s cells in order to exert therapeutic effects. Current mRNA delivery is commonly using lipid-based or viral-based materials as nanocarriers. Their applications are limited due to immunogenicity and storage concerns. This study aims to explore the potential of peptide-mRNA delivery, by looking into the transfection efficiency and the stability based on a previously discovered short peptide, H10 (Fmoc-PKKKRKVG-HHHH-C12-CStBuHHCStBuCStBuHHCStBuGSPHHD-amide). Ultrasound-Triggered Drug/Gene Release from Self-Assembling Carriers Supervisor: CHAU Ying / CBE Student: SI Mingxuan / BIEN Course: UROP1000, Summer Artificial membraneless organelles (AMOs) have been constructed by the self-assembly of polypeptide, called intrinsically disordered proteins (IDPs). By liquid-liquid phase separation, droplets are formed. These droplets are able to contain DNAs, RNAs, and proteins and are sensitive to environmental stimuli such as temperature, pH, or ultrasound trigger. We investigated the partition of DNA inside the droplets by using Pico-green, a DNA quantification assay. Moreover, we demonstrated the effect of using ultrasound on the droplet size and the potential that the droplets release their cargos. Droplets with smaller sizes are found in the ultrasound-treated samples. We anticipate releasing the cargos inside the droplets in a controllable manner using ultrasound. Cloud-Based Problem and Solution Sharing Platform for Injection Molding Supervisor: GAO Furong / CBE Student: CHEN Duoli / CENG Course: UROP1100, Fall The unbalance in talents and demands is a serious problem encountered in injection molding industry. Therefore, this project aims at developing a cloud platform which can be utilized by users in the form of WeChat mini-app. Embedded in the most popular and convenient media in China of WeChat, the platform enables the recommendation and matching of technicians and factories through quests and contributes to easier communication in between, saving much trouble and enhancing the efficiency of the industry. It also provides an approach of data collection for the future investigation and comprehensive learning on defects fixing and injection molding machine adjustment. 81 School of Engineering Department of Chemical and Biological Engineering Designing of the Chemical Process for Polyethylene Furanoate-Based Materials Supervisor: HU Xijun / CBE Student: YANG Jianbo / CENG Course: UROP1100, Spring In today’s logistics and construction industry, wooden materials are in huge demand. There is a significant number of wood waste generated from those used pallets and boxes, which has occupied up to 10% of landfill capacity in Hong Kong. To reduce the waste of materials, these recyclable wood wastes need certain chemical methods which can be transform them into other useful organic substances or products. At the same time, a potential opportunity to replace petroleum-based materials lies ahead of wood materials since they are abundant around the world and have a high-strength hierarchical structure. Thus, the project aims to synthesize a transparent wood material and see if we can functionalize it by chemical modifications. Designing of the Chemical Process for Polyethylene Furanoate-Based Materials Supervisor: HU Xijun / CBE Student: YANG Jianbo / CENG Course: UROP2100, Summer A significant number of wood waste occupying up to 10% of landfill capacity in Hong Kong is generated through the dispose of wood pallets and furniture. To increase the utilization rate of the material and ease the pressure on landfill space, an applicable chemical method to transform wood waste into useful organic substrates or products is expected. Wood is a kind of natural macromolecular material with a complex hierarchical structure. It is composed of cellulose, hemicellulose, and lignin. These components can be extracted through chemical methods and used in different fields after different modifications. The rest of the paper will discuss the experiments of these components and their potential uses. Synthesized Porous Materials For Water Purification Supervisor: KIM Yoonseob / CBE Student: LAI Chit Him / CEEV Course: UROP1100, Fall Metal-Organic Frameworks (MOF) have been an emerging technology in various industries. Its cage-like, hollow, and highly porous structure gives MOF extremely large surface area, tunability and flexibility. These features enable MOF to have a wide application such as water splitting, gas separation and electrochemical energy storage. Different combinations of MOFs are investigated to maintain its stable structure in different environments and usage. In this project, the removal of metal ions in water will be focused by testing the kinetics and capacity of Zr4+ based MOFs, which is highly stable in water. It is shown that the ratio of nodes and ligands does not affect the capacity whereas OH group shows a weaker removal than thiol group due to its weaker electronegativity. 82 School of Engineering Department of Chemical and Biological Engineering Influence of Random Base in Poly(A) Tail to the Function of Mammalian mRNA Supervisor: KUANG Becki Yi / CBE Student: SHI Naiyu / BIEN Course: UROP2100, Spring Messenger ribonucleic acid (mRNA) is a single-stranded RNA. It delivers the genetic information of the DNA to the cytoplasm to synthesize proteins. To protect mRNA from degradation, a poly(A) tail is added at the 3’ end. Despite the name suggests it is a string of adenines, random insertions of guanine (G), cytosine (C) and uracil (U) inside poly(A) tails have been found. These insertions seem to have the capability of affecting the half-life and the translation efficiency of the mRNA, however, whether these insertions are the reason for the changes and how they influence the functions and properties of mRNA are not known. This project aims to investigate the influence of insertions of G and C in poly(A) tails of mammalian mRNA. RNAs with different poly(A) tails will be made in vitro and transfected into mammalian cells. Flow cytometry will be applied to analyze the transfection and protein expression rate. The Construction of Dual-Functional Synthetic RNA Control Unit Supervisor: KUANG Becki Yi / CBE Student: TSE Richard Chi Hong / BCB Course: UROP1000, Summer In this project, we have demonstrated that by using RNA-only delivery method, we could enable direct activation and repression on the translation of mRNA, our system could minimise the components needed, and it also expands the current regulatory components as in the field of synthetic biology and as a safer alternative in different medical and therapeutic applications. We demonstrated that RBP-aptamer pair PP7CP could repress translation of mRNA with a translational active 5’-cap analog with superior translational activity, whereas activates translation of mRNA with a translational inactive 5’-cap. Improving Data Analysis Methods in Proteomics Supervisor: LAM Henry Hei Ning / CBE Student: RODRIGUEZ GUTIERREZ Mauricio Antonio / MATH-PM Course: UROP1100, Fall This report will first describe common techniques applied in the field of Proteomics for the identification of peptides from mass spectrometry data. Afterwards, describe a methodology to identify features on mass spectrometry scans between matches for tryptic cut human peptides depending on their Cterminal (Lysine or Arginine). The two techniques used for analysing the features were targetdecoy database search, and the data analysed is a draft map of the human proteome submitted to the PRoteomics IDEntification Database (PRIDE). The report hopes to investigate any possible differences between the ease of identification for fragmentation patterns depending on the type of tryptic cut site. Finally, it will present the resulting data, which was filtered, processed and presented using Python library Pyteomics for the handling of .pep.xml files. 83 School of Engineering Department of Chemical and Biological Engineering Improving Data Analysis Methods in Proteomics Supervisor: LAM Henry Hei Ning / CBE Student: LEE Jooran / BIEN Course: UROP1100, Spring Peptides from digested proteins are chromatographically separated and fragmented in the mass spectrometer, generating fragment ion spectra that can be used to deduce the corresponding peptide sequences. These Peptide Spectra Matches (PSMs) in turn allow the identification of the proteins in the biological sample. The PSMs, however, are not always correct. Using several well-known machine learning algorithms to separate correct PSMs from the incorrect PSMs was the main purpose of the project. It was done by comparing their accuracy, performance, and time taken for the classification. In addition to that, the feature importance is calculated from each machine learning algorithm and the common important features were extracted that can potentially be used to build a more efficient model for data analysis on proteomics. Second part of this research was based literature review on evaluating computational methods and error rate estimation procedures for peptide and protein identification in shotgun proteomics. Improving Data Analysis Methods in Proteomics Supervisor: LAM Henry Hei Ning / CBE Student: ISLAM Kazi Anisha / CENG Course: UROP1000, Summer The protein modifications of bacteria have not been extensively researched and hence, an investigation of their complexity is worth being carried out. In this project, an open search analysis of E.coli peptide MS2 spectra was conducted. Relatively low numbers of peptide-spectrum matches (PSM) with uncommon modifications (hydroxylation of phenylalanine and tyrosine, di/oxidation of tryptophan, cysteine →dehydroalanine) were identified. Due to the limitations of data dependent acquisition method and a potentially high false negative rate, PSMs of the peptides with the uncommon modifications are observed in low abundance. While oxidation and deoxidation of tryptophan hydroxylation of tyrosine are posttranslational modifications, cysteine →dehydroalanine and hydroxylation of phenylalanine are chemical artifacts. In general, it can be observed that untreated E.coli is less likely to produce proteins with uncommon modifications. Development of Valuable Fine Chemical Products from Wooden Waste by Heterogeneous Catalytic Pyrolysis Supervisor: LAM Leung Yuk Frank / CBE Co-supervisor: HU Xijun / CBE Student: LIEM Albert / SUSEE Course: UROP1100, Fall As a metropolitan and one of the most densely populated places in the world, land prices in Hong Kong is exorbitantly high. With limited land supply and increasing demand for space, comprehensive and strategic planning is required to provide much needed capacity for waste disposal in Hong Kong. With the rapid growth in the logistics and construction sectors, those factors have led to a large demand for wood pallets for easy transportation. Typically, pine pallets are preferred due to its low cost and it cheap enough that pine pallets are often only used once before being abandoned at the end of the supply chain, consequently ending up in landfills and contributing to wood waste that occupies 8-10% of the landfill capacity in Hong Kong. Thus, this project aims to develop specially designed metallic-based catalyst to transform wooden waste to more useful organic compounds such as pyroligneous acid and bio-oil. 84 School of Engineering Department of Chemical and Biological Engineering Development of Shower Pills (Instead of Gel or Soap) Supervisor: LUO Zhengtang and LIU Y.S. Marshal / CBE Student: TSANG Hoi Lam Tammy / CHEM WONG Yung Tat Vincent / SENG Course: UROP1000, Summer UROP1000, Summer This report will focus on the development of shower pills including the selection of chemical ingredients based on literature review, the formation of shower pills based on different trials, the shortcomings and merits of the shower pills, the problem we face and the improvements we can make. Overall, the development of shower pills is not very successful. Only two out of eight trials can form a pill successfully and the pill becomes sticky when pressure is applied to it. This is because of the chemicals' characteristics, which leads to a paste-like texture performed by the shower pills and a long dissolving time. We believe that a change in chemicals and improvement in equipment can ease the problems we face. Advanced Catalysts for Water Splitting and Hydrogen Production Supervisor: SHAO Minhua / CBE Student: KIM Hyeongwoo / CENG Course: UROP1100, Fall As the movement for the decrease in carbon footprint rises, alternative fuels for various applications are available and investigated in modern society. Electrochemical hydrogen evolution reaction (HER) is considered a renewable and feasible way of generating hydrogen fuels for the future. Various types of catalysts used for the reaction have been extensively discussed for previous decades. Pt/Ru bimetallic catalyst is well known for its promising performance, but the effects of annealing temperature, which is used for pyrolysis of catalyst production, on HER performance has not been discussed in the literature. In this project, Pt/Ru bimetallic catalysts are produced with different annealing temperatures, and its electrochemical active sites and HER performances are discussed through cyclic voltammetry and linear sweep voltammetry. Advanced Catalysts for Water Splitting and Hydrogen Production Supervisor: SHAO Minhua / CBE Student: YANG Jianbo / CENG Course: UROP1100, Fall Carbon dioxide is one of the greenhouse gases causing global warming due to its ability to absorb infrared light from the Earth and return some of it, causing the Earth to heat up. Researchers nowadays pay more attention to electrochemical reduction of carbon dioxide for solving problems like means of energy storage or environmental issues. Advanced catalysts design is a popular research topic since different compositions and structures will have different effects on products. Therefore, finding an economical and efficient catalyst is significant to current research, even to industrial production. In this program, my supervisor and I focused on synthesizing Pd/C nanoporous catalysts loaded with pyridine-derivatives and analyzing their performances under a series of voltages in electrochemical reduction reaction. 85 School of Engineering Department of Chemical and Biological Engineering Cathode Materials for Lithium-Ion Batteries Supervisor: SHAO Minhua / CBE Student: KIM Hyeongwoo / CENG Course: UROP1100, Spring Among many options, high voltage cathode materials are considered a vital element for the next-generation Li-ion battery. How to improve the cycling stability of LiNi0.8Co0.1Mn0.1O2 (NCM811) is a hot topic because of its outstanding capacity density, good rate performance but unfavorable cycling stability nature. The solid electrolyte interface (SEI) modification through electrolyte additives is one way of improving the performance of the cell, and fluoroethylene carbonate (FEC) is frequently used as a co-solvent. Cells assembled with NCM811 cathode, Li metal anode, and FEC-based electrolyte are subjected to charging and discharging cycles to observe the capacity retention after 100 cycles. A color-changing phenomenon related to FEC-based electrolytes is also discussed. Developing Genetically Encoded Molecular Tools for Controlled Intracellular Phase Separation Supervisor: SUN Fei / CBE Student: LO Shin Dawn / BIEN Course: UROP1100, Summer CarHc is part of a photoreceptor protein that can be split into two segments known as CarHcN and CarHcC, which contains the cobalamin-binding domain. Since CarHc is a split protein, it has the potential to be controlled by researchers via induction of its components to reassemble and even disassemble. This property can be applied to study proteins, control molecular function, and assemble protein architecture. In this study, the cobalamin-bindingcapacity and subsequent photosensitivity of CarHcC was investigated. It was found that CarHcC alone was indeed capable of binding with cobalamins cyanocobalamin (CNB12), adenosylcobalamin (AdoB12), and methylcobalamin (MeB12); the protein was even photosensitive when bound to AdoB12. This discovery confirms that the binding domain is located on the C-terminus split protein component. This discovery has implications for the possibility of genetic modification of CarHc – specifically, the CarHcN component – to tailor the protein for specific applications. Analyzing the Possible Water-Soluble Proteins for Deep-Tissue and Cancer-Targeting in Vivo Imaging Using a Photoacoustic Imaging System Supervisor: WONG Tsz Wai / CBE Student: CHENG Leong / CHEM Course: UROP1100, Summer To overcome the disadvantages associated with cancer-diagnosis methods commonly used today, including low molecular specificity from ultrasound imaging, highenergy radiation from X-rays, and gadolinium-deposition nature of MRI, the photoacoustic imaging system is proposed. This modality allows for non-invasive and deep-tissue in vivo imaging through ultrasound waves generated by the thermoelastic expansion that is induced by light energy absorbed by tissues. However, in order to produce high resolution photoacoustic images, a highly specific contrast agent that satisfies several criteria is required; some of these critical criteria include aqueous stability, biocompatibility, and redshift. Thus, this report investigates several potential candidates for the contrast agent and selects the optimal one using the aforementioned criteria. 86 School of Engineering Department of Chemical and Biological Engineering Analyzing the Possible Water-Soluble Proteins for Deep-Tissue and Cancer-Targeting in Vivo Imaging Using a Photoacoustic Imaging System Supervisor: WONG Tsz Wai / CBE Student: LAI Ka Wai / SENG Course: UROP1100, Summer Early diagnosis of cancer can greatly increase the chance of curing the disease. Existing imaging modalities for cancer detection may be too costly and time-consuming. They also have disadvantages like the low molecular specificity and penetration depth. A photoacoustic imaging system can overcome the limitations and improve image quality. This study will investigate some possible contrast agents. The agent used should be biodegradable, cancer-targeting, and having a versatile molecular design, a near-infra-red absorption peak. The contrast agent can also potentially perform photodynamic therapy which makes the imaging system a cancer-curing device simultaneously. The system should apply single-wavelength differential imaging to improve the image quality. An optimal candidate for the protein in the contrast agent is BphP1. Development of a Deep Learning Approach for Transferring Gray-Scale Images Acquired from Photoacoustic Imaging to Histological-Stained Images Supervisor: WONG Tsz Wai / CBE Student: SHI Naiyu / BIEN Course: UROP1100, Fall Histology staining has been universally used in hospitals for pathologists to give diagnoses. However, the process of histology staining is complicated and takes a long time before the skilled pathologists could make a judgement. In this work, we explore the use of deep learning methods to transfer the gray-scale images obtained by microscopic imaging technique into histological-stained images. Deep learning, as a subfield of machine learning methods, is based on algorithms inspired by the structure and function of the brain, named artificial neural networks. We specifically look into a model of Generative Adversarial Network (GAN) called CycleGAN, study the code of it and also its capabilities. Development of a Deep Learning Approach for Transferring Gray-Scale Images Acquired from Photoacoustic Imaging to Histological-Stained Images Supervisor: WONG Tsz Wai / CBE Student: LO Shin Dawn / BIEN Course: UROP1100, Spring Surgery is often used to treat local-stage breast cancer (American Cancer Society, 2020; Wong, et al., 2017; Wong, et al., 2018); the current industry practise for evaluating the success of a surgery is to stain tissues post-surgery. In the case that a tumour is not completely removed, patients need to undergo follow-up surgeries, which is timely and costly. Photoacoustic microscopy is a promising imaging technique that can identify cancerous cells in vivo during surgery, reducing the need for follow-up surgeries. To improve the usability of PAT, deep learning is applied to translate PAT images into histological-like images that surgeons are accustomed to viewing. The following paper evaluates the use of deep learning for image translation and suggests other computational methods that could improve the functionality of PAT. 87 School of Engineering Department of Chemical and Biological Engineering Development of a Deep Learning Approach for Transferring Gray-Scale Images Acquired from Photoacoustic Imaging to Histological-Stained Images Supervisor: WONG Tsz Wai / CBE Student: KOBASHI Minato / BIEN Course: UROP1100, Summer We investigate supervised deep learning as a method to enhance resolution of tissue images. High-resolution microscopic tissue images are often difficult to obtain due to limited penetration depth and requiring expensive imaging devices. Conditional adversarial networks can be trained to learn a loss function on its own through a conditional setting which makes this network a generalized solution for problems that require complex loss functions. We borrow the idea from a pix2pix software and introduced two main components to our network, U-Net generator and PatchGAN discriminator as our adversarial network. In this report, we have shown that our conditional adversarial networks are capable of generating higher resolution tissue images. Mask but not the Least Supervisor: YEUNG King Lun / CBE Student: BALAJI Kavya Geetha / DSCT KANDWAL Inika / EVMT Course: UROP1100, Spring UROP1100, Spring The COVID-19 pandemic has not only affected and hindered our daily lives, or the environment today, but has asked us to seriously consider the environmental impacts of our lifestyle choices. A prominent issue at the moment is the unattended disposal of face masks, around 129 billion face masks every month. These discarded face masks pose several environmental, social and health issues. This paper outlines a detailed process of collecting, disinfecting and repurposing disposable masks on campus to create products of high-value and function as well as using them to raise awareness around the importance of proper mask disposal. Mask but not the Least Supervisor: YEUNG King Lun / CBE Student: BALAJI Kavya Geetha / DSCT KANDWAL Inika / EVMT Course: UROP2100, Summer UROP2100, Summer The COVID-19 pandemic has not only affected and hindered our daily lives, or the environment today, but has asked us to seriously consider the environmental impacts of our lifestyle choices. A prominent issue at the moment is the unattended disposal of face masks, around 129 billion face masks every month. This paper explores the creation of composite prototypes using shredded mask waste and other waste materials (paper, natural fibres etc) to create load-bearing, water resistant and aesthetic material. We aim to create products of high-value and function as well as using them to raise awareness around the importance of proper mask disposal. 88 School of Engineering Department of Civil and Environmental Engineering Department of Civil and Environmental Engineering Development of a Red Tide Early Warning System for Fisheries Management Supervisor: CHAN Shu Ning / CIVL Co-supervisor: LEE Joseph Hun-wei / CIVL Student: YAO Xihan / EVMT Course: UROP2100, Fall Red tide is the discoloration of water caused by excessive algal growth under favorable environmental conditions. Chlorophyll-a concentration (Chl-a) is an important water quality parameter representing the algal biomass in water. Conventionally, Chl-a is measured with manual water sampling and laboratory analysis, which is labor-intensive and time-consuming, resulting in delayed response to environmental disasters. This project aims at developing an efficient method for seawater quality monitoring and red tide forecasting. Using spectral reflectance data from multispectral drone images and simultaneously measured Chl-a in a fish farm in Tolo Harbor, New Territories, Hong Kong, a correlation study was conducted between different wavelengths’ reflectance and Chl-a to develop an algorithm for drone-based Chl-a monitoring. The correlation study consists of three stages - starting with single variable regression, then multivariable regression, and Principal Component Analysis (PCA) with SPSS software to test for the best-fit correlations. Improving Direct Filtration Performance of Water Treatment Works Supervisor: CHAN Shu Ning / CIVL Co-supervisor: LEE Joseph Hun-wei / CIVL Student: LEE Kin Ming / CIGBM Course: UROP1100, Fall Algal blooms can occur in freshwater reservoirs under favourable temperature conditions and sufficient nutrient level, and can lead to a deterioration of source water quality. In this report, the techniques in removing algae in reservoirs and water treatment works are reviewed, with a focus on water filtration technology. Field measurements of turbidity, temperature, pH, water level and head loss were taken in Tsuen Wan Water Treatment Works to study the performance of direct filtration. The collected results illustrate overall coherence with general filter bed design theories from various literature, such as the range of influent turbidity and the trend of head loss increase. Engineering Properties of Soil Amended by Hydrochar, a New Resource Derived from Food Wastes Supervisor: LEUNG Anthony / CIVL Student: HUI Yee Lam Elim / SENG Course: UROP1000, Summer Hydrochar is a type of char produced through hydrothermal carbonization process, where wet biomass is heated in the presence of water under pressure at a temperature of 150-375°C. Using coffee grounds as feedstock, the main objective of this project is to investigate the applicability of coffee grounds and coffee ground-based hydrochar as soil amendment in completely decomposed granite (a local soil) through evaluating the shear strength in terms of cohesion and friction angle. The varied results showed that coffee grounds and coffee ground-based hydrochar might be possible soil amendments. However, effective utilization would require further research and investigation. 89 School of Engineering Department of Civil and Environmental Engineering Engineering Properties of Soil Amended by Hydrochar, a New Resource Derived from Food Wastes Supervisor: LEUNG Anthony / CIVL Student: LEUNG Chun Yu / CIVL Course: UROP1000, Summer Geotechnical engineering is primarily concerned with determining the strength and deformation characteristics of soil. In geotechnics, clay, silt, sand, rock, and snow are essential materials. Soil and rock mechanics, geophysics, hydrogeology, and related sciences such as geology are all part of geotechnical engineering. Civil engineering includes geotechnical engineering and engineering geology. While soil mechanics is one of the aspects we have got in touch with frequently, especially in Hong Kong. With mountainous terrain, it is common to see different natural and man-made slope in the city. These slopes need to be maintained in order to keep its stability and ecological value. Recently, engineers are trying to develop a mixture of soil and hydrochar to improve soil property. In this project, the soil behaviours mixed with coffee grounds and biochar burnt from coffee ground were investigated through a series of direct-shear tests. Engineering Properties of Soil Amended by Hydrochar, a New Resource Derived from Food Wastes Supervisor: LEUNG Anthony / CIVL Student: MAK Cheuk Yiu / IPO Course: UROP1000, Summer Hydrochar is a product of the hydrothermal carbonization of biomass, which is obtained in the form of slurry, a mixture of liquid and solid. It could be produced with either dry or wet biomass, with water added during the hydrothermal carbonization process. This project considers the effects on geomechanical properties when hydrochar derived from coffee ground is mixed into completely decomposed granite soil (CDG), as soil amendment is known to be an important application of hydrochar. Apart from the hydrochar, similar experiments were carried out with its feedstock. This project compares their effects on soil shear strength, which appeared to be rather diverse in terms of soil cohesion and friction angle. Engineering Properties of Soil Amended by Hydrochar, a New Resource Derived from Food Wastes Supervisor: LEUNG Anthony / CIVL Student: WONG Tsz Fung / SENG Course: UROP1000, Summer Geotechnical engineering is one of the civil engineering disciplines that involves detailed investigation in the behaviour of soil and rock, design of geotechnics work like slope stability, foundation piling, material testing, risk assessments of mass movement hazards. Coping into our urban living, the coexistence of both the concrete and natural jungles all around proves the significance of geotechnical engineering, and its value to the development of infrastructure construction and the community safety. To better enhance our living standards and safety, engineers are developing a mixture of soil and ameliorated materials - biochar, to improve soil property. In this project, the shear strength of soil was measured by comparing the behaviour of coffee grounds and its derived hydrochar in saturated conditions through a series of direct-shear tests. 90 School of Engineering Department of Civil and Environmental Engineering Shear Strength Properties of Hydrophobic Soil Supervisor: LEUNG Anthony / CIVL Student: LI Sau Long / CIVL Course: UROP1100, Fall Hydrophobic sand, which has been utilised to minimise rainfall infiltration, is investigated for its possible use in slope stabilisation. However, how the application of the silicone coating may affect the mechanical behaviour of hydrophobic soil including the shearing properties has rarely been investigated. The purpose of this study is to design and develop a new apparatus, which can monitor any pore water pressure changes during a direct shear test and can determine the effects of suction on the shearing behaviour of hydrophobic soil. The hydrophobicity in sands can be attained by artificially applying a layer of silicone coating, polydimethylsiloxane (PDMS). Various mass ratios of dimethyldichlorosilane (DMDCS) is mixed with Toyoura sand to produce PDMS coatings by silanisation. In this project, a preliminary direct shear test, 25 and 50kPa tests and a validation test on only untreated and unsaturated samples were conducted. Results gave reasonable trends with increasing normal stress but formed shear stress peaks less prominent than expected. In the future, the shear strength and dilation behaviour of treated samples will be studied through suction-monitored direct shear tests. Engineering Properties of Soil Amended by Hydrochar, a New Resource Derived from Food Wastes Supervisor: LEUNG Anthony / CIVL Student: WAN Ngo Yeung / CIVL Course: UROP1100, Fall Hydrochar is charcoal derived from biomass and has been used for soil amendment and modifying the soil properties. However, effects of hydrochar on the geotechnical properties including shear strength are unknown. This project investigates the effects of biochar on the shear strength properties of a local soil, completely decomposed gravel (CDG). Direct-shear tests were conducted to compacted CDG samples in order to measure soil’s shear stress, vertical displacement, and horizontal displacement. It is found that hydrochar increases the vertical displacement of dry CDG but decrease that of moist CDG (at optimum water content). Hydrochar has minimal effect on the shear stress of dry CDG but decrease that of moist CDG. Hydrochar decreases the friction angle of both dry and moist CDG. Recyclable Hydrogel-Based Brick for Construction on Mars: A Feasibility Study Supervisor: QIU Jishen / CIVL Student: NGAN Man Cheuk / CVGBM Course: UROP2100, Fall This study aims at investigating the feasibility of recycling hydrogel-based bricks and the impact to its mechanical properties after recycle process. Our research object is Gelatin as it provides high recyclability with little requirement of additional chemical agents and environment condition. Compression tests and SEM analysis are conducted to examine the impact of recycling bricks to the stress, elastic modulus and failure modes of the hydrogel-sand composites. The bricks recycled by heating with water only can attain an elastic modulus of 177.4MPa and a compressive strength of 6.45MPa, which is about 70% of the original strength before recycling. 91 School of Engineering Department of Civil and Environmental Engineering Recyclable Hydrogel-Based Brick for Construction on Mars: A Feasibility Study Supervisor: QIU Jishen / CIVL Student: NGAN Man Cheuk / CIGBM Course: UROP3100, Spring This study aims at investigating the efficiency of recycling hydrogel from the hydrogel-based bricks and the impact to the gel’s mechanical properties after recycling. Our research object is Gelatin as it provides high recyclability with little requirement of additional chemical agents and environment condition. Compression tests and SEM analysis are conducted to examine the impact of recycling bricks to the stress, elastic modulus and failure modes of the bricks. The gel extraction process by heating with water can obtain 80% of the original gel adhered on the surface of ceramic beads and new bricks made with gel extracted can maintain a compressive strength of 6.55MPa, which is about 63% of the original strength before recycling. DESR (Data-Enabled Scalable Research) Lab - Incubator of Creative Learners and Hub of Transformation Supervisor: WANG Yu-Hsing / CIVL Student: LEE Jae Yeol / MATH-AM Course: UROP1100, Spring Risk assessments of trees has been a crucial challenge, especially in Hong Kong where the city is densely populated and frequently faces typhoons. Fatal accidents regarding tree failures have constantly been reported, particularly during the super typhoon Mangkhut. To check the internal state and structure of the tree, sonic tomograph and electrical tomograph have been used. It is a vital technique in measuring the damage of the tree and whether tree failures will happen or not. Tree pulling test is also a renowned experiment method of inspecting tree motions that occur under varying weather conditions. The details of the technique LiDAR in extracting tree data points will be discussed, as well as the procedures of the tomography. Internet of Tree Things (IOTT) Supervisor: WANG Yu-Hsing / CIVL Student: CHAN Chun Yu / SENG Course: UROP1100, Fall The first part of this report will focus on the introduction of IOTT. This report will first introduce IOTT and compare it with LiDAR. LiDAR is an alternative method to measure the tree movement. This comparison can reflect the merits and drawbacks of IOTT. The second part of the report will focus on tree selection. This report will share some special tree conditions that should be cautious. The third part of this report will focus on sensor manufacture. This report will include the procedure and difficulties of building a printed circuit board for a sensor and introduce how NeoDen 4 can advance the manufacture. Finally, the fourth part of the report will focus on the data analysis after receiving raw data. This report will introduce a method to calibrate those raw data and share some results. 92 School of Engineering Department of Civil and Environmental Engineering Internet of Tree Things (IOTT) Supervisor: WANG Yu-Hsing / CIVL Student: CHAN Chun Yu / CIEV Course: UROP2100, Spring The first part of this report will focus on the isolating tree project that started the last semester. This report will first introduce Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and show the result using this algorithm. The second part of the report will focus on Picus Sonic Tomograph and TreeTronic in tree diagnosis. This report will share the procedure of performing tree diagnosis and show the results of special tree conditions. The third part of this report will focus on the determination of the tree’s elastic modulus. This report will include the procedure, problems, and advice of the test. Finally, the fourth part of the report will focus on the tree pulling tests. This report will introduce the procedure of the test and the result. You will see a Hoopoe Supervisor: WANG Yu-Hsing / CIVL Student: KWOK Martin Ying Kit / CIVL Course: UROP1000, Summer “You will see a Hoopoe” is a sustainable smart campus project using innovative technology and AI-enabled camera to identify birds in the HKUST campus. In fact, birds play a very important role in the nature while HKUST campus is a perfect location for students to explore the nature. It is discovered that there are more than 46 species of birds live around the campus and we are trying to recognize these birds with the devices. The project is aiming at promoting the environmental awareness of bird protection through the green AI camera which is most advanced bird-watching system in Hong Kong. Coupling Sulfur-Driven Denitrification with Anammox for Wastewater Nitrogen Removal Supervisor: WU Di / CIVL Co-supervisor: CHEN Guanghao / CIVL Student: CHUI Wai Yuen Ivan / CIVL Course: UROP2100, Fall This study aimed to investigate a new sulfur-cycle biotechnology for wastewater treatment. The experiments were carried out to study the feasibility of three types of anaerobic sulfate reducing reactors, namely, (a) sulfate reducing UASB reactor without mechanical stirring, (b) fluidized granular sludge bed using 0.074 –0.125 mm micro sand as growth media, and (c) fluidized granular sludge bed using 0.07 – 1.0 mm wood charcoal as growth media. These three reactors were seeded with anaerobic sludge from a food waste treatment plant. The influent COD was 500 mg/L, and the hydraulic retention time was 20 h. The results indicated that the sulfate reducing UASB reactor without mechanical stirring was unable to operate due to the lack of biogas formation for internal mixing. Fluidized bed reactor using micro-sand was also unable to operate smoothly due to the high density of sand. Fluidized bed reactor using wood charcoal was smoothly operated for 3 months. The COD removal achieved 55% and there was a significant growth in bacteria. Moreover, with charcoal, the upflow velocity in anaerobic reactor could be increased from 1m/h (for traditional UASB reactor) to 6m/h. In future, FISH test should be done to further study the communities of bacteria proliferated on the charcoal media; and in addition, more experiments should be done to figure out the optimum size range of charcoal. 93 School of Engineering Department of Civil and Environmental Engineering Coupling Sulfur-Driven Denitrification with Anammox for Wastewater Nitrogen Removal Supervisor: WU Di / CIVL Co-supervisor: CHEN Guanghao / CIVL Student: CHUI Wai Yuen Ivan / CIVL Course: UROP3100, Spring This study aimed to investigate a new sulfur-cycle biotechnology for wastewater treatment. An expanded/fluidized granular sludge bed reactor, using wood charcoal granules as the carrier, was being investigated for 220 days. The influent chemical oxygen demand (COD) was 300 mg/L, and the hydraulic retention time (based on expanded bed volume) was 20 hours. An overall COD removal efficiency of 47% obtained and the volumetric COD loading for expanded bed height was 0.40 kg/m3 /day. Sludge characterization was also determined. A new method was developed based on Wang’s Method to separate the organic-COD and the sulfide-COD from the total soluble COD in the effluent. Optimized Design of Urban Neighborhoods for Environmental Sustainability: The Effectiveness of Different Strategies Supervisor: YANG Jiachuan / CIVL Student: LI Sau Long / CIVL Course: UROP1100, Summer Global climate change and its related phenomenon leading to extreme temperatures have become an increasing threat to sustainability in cities. A variety of novel engineering materials, building technology and urban planning methods have been proposed in recent years as strategies to optimise thermal comfort and building energy efficiency. This project aims to review the potential strategies in the literature, comparing their impacts on urban canyons and the respective building energy efficiencies through numerical modelling (Urban Canopy Model). The effects of different strategies are tested through tweaking variables in the model simulation, providing Pareto fronts between thermal comfort and building efficiency and thus identify potential conflicts between the two. In this study, the effects of the strategies on the urban climate will be assessed. Population Exposure to Hazardous Weather under Climate Change Supervisor: YANG Jiachuan / CIVL Student: LEE Lok Hei / CIVL Course: UROP1000, Summer Occurrence of heat wave can impose serious impacts on human health, agricultural activity, the environment, and ecosystems. In view of climate change and resultant weather alteration, it is important to study how the future weather conditions may be different from the present one and prepare for it. In this study, we decided to place our focus on analyzing the future heat wave pattern and its possible effects in the Northern Hemisphere. We found that across the Norther Hemisphere, Africa will experience the greatest impacts under climate change. By studying how the maximum temperature and heat wave duration vary in the future, we are able to estimate the influence of extreme weather. 94 School of Engineering Department of Civil and Environmental Engineering Applying 'Lean' Principles to Improve Construction Productivity Supervisor: ZHANG Xueqing / CIVL Student: YIN Junyu / CIGBM Course: UROP1000, Summer This report summarises all the work that I have done for the research project ‘Applying 'Lean' Principles to Improve Construction Productivity’ conducted by Prof. Zhang, Xueqing. In these six weeks of UROP project, I performed my work under the instruction of Dr. Ahmed, Rana Rab Nawaz. I have read serval literatures regarding the topic of Lean construction, value stream mapping, waste management, and graphical user interfaces and prepared summaries of them. Also, I have developed a functional user interface using Excel VBA for estimating different costs regarding the disposal of construction waste. This app will be useful for future research work: a short case study with real data. 95 School of Engineering Department of Computer Science and Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: PRINTZ Maximilian / COGBM Course: UROP1100, Fall This report specifies my participation in designing a highly accurate classification-based proximity detection model, which is a winning approach in an international open challenge (TC4TL challenge). After introducing the structure of our proximity detection model, this report describes 3 major experiments conducted after the competition to further learn about what factors drive a successful proximity detection model. Brief descriptions of each experiment are given, and their results are explained through comprehensive charts and diagrams. Given the recent major challenges caused by the Covid-19 pandemic, developing an effective and efficient proximity detection model for contact tracing is critical. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: WU Qi / COSC Course: UROP1100, Fall Among the many indoor localization methods, wifi-based indoor localization is one of the most popular ones. In this article, I will try to post my understanding of wifi-based indoor localization and some research data. Since I was not on campus, relevant activities are conducted off-lab. I spent the first few week on relevant papers and learning about artificial intelligence after discussing with Amy. Considering I have no background knowledge, I started with the simpler ones, ideally to review wideep’s experiments. The first part of the article is the understanding of the concepts related to wifi-based indoor localization, the second part is mainly some data, and the third part is my feeling of taking the UROP for the first time. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: ZHANG Weiwen / DSCT Course: UROP1100, Fall Crowd counting has been a hot topic with enormous potential in real-world application. Multi-camera crowd counting is one of branches of crowds counting and its general approach is in field of computer vision and deep learning. This project is based on the approach that to train a CNN model to generate density maps with synthetic image datasets and count the number of people by integrating on a density map. And this report is mainly concentrating on proposing a novel method to generate synthetic images in Unity engine, in which we can add more variety into the dataset to imitate crowds in real-world scenario. The initial result and unity project can be seen on the website: https://github.com/KevynUtopia/multi-camera-crowdcounting_UROP. 96 School of Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: CHIU Ka Ho / COSC Course: UROP2100, Fall As crowd-source WiFi indoor fingerprinting has its unique importance in applications such as monitoring people’s locations and indoor localization, related methods are worth studying. This report would focus on its usage in indoor localization, introduce one proposed method and focus more on improvement details. Wi-Fi fingerprinting has two phases, survey and query, but this report would assume that the survey phase is done and some labeled locations with exact coordinates and RSSI vectors are obtained. The focus would be on the second phase, which is to get a model for predicting the target position based on those obtained labels. The model would be based on the similarities, or proximities between nodes, and is divided into two parts, the first part would perform network embedding for feature extraction, The second part would be a spring-like system with carefully designed loss to enforce various constraints to further project the nodes from lower-dimensional embedding space to 2 dimensional physical space. This report would focus more on the second phase of the project, and emphasize a number of implementations to improve the final results and some outcomes of trials. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: JIANG Tianrui / COMP Course: UROP2100, Fall With the rapid development of modern traffic system, the data collected has become much more diverse and abundant, providing the potential for deeper and more precise analysis, prediction, as well as upgraded recommendation system. In particular, the demand of flow data is currently a hot topic, as there remains a large room for improvement on current solutions. An important category of methods are based on origin-destination pairs, which basically have better performance than zoon-based methods with the cost of additional complexity. Inspired by the prevalence of attention mechanism in the area of natural language processing and computer vision, this project seeks to discover a way that integrates it into origin-destination prediction. My work involves researching relevant works and participating in the implementation of experiments for control-group methods. The proposed method is not finalized yet, and needs further enhancement. 97 School of Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: PENG Zhuoxuan / COSC Course: UROP2100, Fall License plate recognition (LPR) is an important computer vision task in transportation and vehicle management. In our project, LPR is performed in real time on Raspberry Pi, a power efficient device. This requires a series of techniques to reduce the computational resources and boost the recognition speed. This semester I first finished the web streaming system, then contributed to the development of processing of pictures and recognition. The major achievement is the implementation of a calibration algorithm by picking out only areas possibly containing license plates. Next I will focus on the connection of different systems in this project. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: XIANG Letian / COMP Course: UROP2100, Fall With the development of digital economy, huge amounts of geographical data that depicts paths from an origin to a destination are created daily. They are generated to businesses like ride-hailing services (for example, Uber and Didi), food delivery services (for example, FoodPanda, DoorDash and Meituan), and express delivery services (for example, FedEx and USPS). Predicting the data for different origin-destination pairs is essential for facilitating these services’ qualities. To find an effective solution to this problem, we need to make use of spatial-temporal series to find the correlations with deep learning techniques. We will also show the result of studies of some model from previous works on traffic prediction from spatial-temporal similarity. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: ZHANG Daofu / COMP Course: UROP2100, Fall In this project, we are aiming to develop a system which can be planted in RPi to automatically recognize license plate information from given images of automobiles. This can help people to manage the car information in the parking lot. There are mainly two steps, first we need to locate the place of the license plate from the given images which are usually taken from cameras in the parking lots. After we obtained the image of the license plate, we used another model to recognize the symbols in it and produce the final result. 98 School of Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: LIANG Shengnan / COMP Course: UROP1100, Spring This report gives an overview of the UROP project-AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data in spring semester 2021. First, it will introduce the project we are mainly focusing on (i.e. Kaggle contest: Indoor Location & Navigation-Identify the position of a smartphone in a shopping mall). Then introduce what relative knowledge was discovered and how to learn them from the perspective of our project and how to combine them with the project. Then two important steps in trajectory estimation will be introduced by comparing and analysis different approaches. In the end, this report will also give a summarize of the whole project, evaluate the performance of the project as well as state what have been learnt from the project and how these contribute to further personal development. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: SZE Kai Tik / COGBM Course: UROP2100, Spring This paper reviews the papers regarding density map application, especially for detection, aiming to find the challenges inside density map regression during the application in reality. In this research, difficulties about density map regression for detection will be concluded initially, and then the combination of detection use and counting use will be discussed later. After showing the challenges behind the density map regression for detection use, different neural network methods for detection will be reviewed and compared with each other so that this paper will help find the pros and cons among different training processes for reference of future research and study. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: ZHANG Weiwen / DSCT Course: UROP2100, Spring Crowd counting has been a hot topic with enormous potential in real-world application. This project is based on the approach that to train a CNN model to generate density maps with synthetic image datasets. Based on the previous work, we generate the dataset from Grand Theft Auto V (GTAV) in this paper by using GCC-CL. Then we implement the domain adaptation to transfer the style of synthetic dataset to real world-world via CycleGAN. This project mainly generates synthetic dataset with GCC-CL and transfers the fake images by CycleGAN with SSIM embedding. And also tried to complete counting task on ShanghaiTech dataset with CSRNet. Source code of this report is available at: https://github.com/KevynUtopia/GTAV_MultiView-SyntheticDataset-DomainAdaptation 99 School of Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: CHIU Ka Ho / COSC Course: UROP3100, Spring WiFi fingerprinting can be of great importance in indoor localization, monitoring objects, and tracking people, which gives value to the study of the method to analyze it. In the previous work during UROP 2100, we have developed a model that is based on network embedding and clustering method, and have performed some experiments on the data collected from the second floor of the main building of HKUST. In this report, we will cover some improvements that we have made to the existing model and would provide more comparisons to illustrate the performance of our model. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: PENG Zhuoxuan / COSC Course: UROP3100, Spring License plate recognition (LPR) is a deeply explored task, but difficult to be implemented on compact computers such as Raspberry Pi. Last semester I implemented a calibration algorithm which can recognize all the areas where a license plate may appear in a fixed scene. To further reduce the computation and accelerate the recognition, it is necessary to compress the recognition model based on neural network. Therefore, I attempted several different model compression techniques and evaluated their effectiveness. Results show that it is possible to accelerate the recognition process by 3 times without much loss of accuracy. In the next step I will focus on the implementation of the IoT system. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: XIANG Letian / COMP Course: UROP3100, Spring In this semester, I continue to work on the traffic forecasting problem using spatial-temporal data analysis. In traffic forecasting, we aim to predict the inflow and outflow of traffic of a region in the coming time interval using the historical spatiotemporal data. While existing methods may not be light and have limited efficiency, we would like to propose a new model called Spatial-temporal Dynamic Context Model (ST-DCM) for traffic forecasting. ST-DCM makes use of context and self-attention. It is evaluated to outperformed the state-of-the-art models in terms of accuracy and efficiency. 100 School of Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: CANDICE Angeline / MATH-CS Course: UROP1000, Summer This report features a localization technology based on the SiFu framework by feeding spot-based and sequence-based signals into a likelihood fusion and employing a particle filter to map indoor location. In addition, it studies the particle filtering in more detail and presents the mathematical formula used to estimate the distribution using particle filter. After a brief introduction to the fusion framework, this paper shifts its focus to debugging a fusion localization engine on a cloud environment. By changing a Singleton class comprising of static pointers into separate objects for each machine, in order to be less computationally intensive and hence more energy efficient. Lastly, this report suggests a real-world application of localization technology in aiding wandering elderlies. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: WIDJAJA Oscar / COGBM Course: UROP1000, Summer Trajectory recovery is the practice of extracting a path from discrete location data, which are present in massive amounts due to advances in sensors technology and Internet of Things. Nevertheless, this task is often difficult since data is usually noisy, sparse, and sporadic. This project implemented 3 models for trajectory recovery: simple linear implementation, and two Random Forest models with different sets of features. Data was first pre-processed with NumPy into Trajectory and Point objects. They were then laid out in a grid and different models using scikit-learn were used to predict simulated missing location data. However, model performance gave unexpected results such as increases in accuracy when data is sparser. Therefore, future work is needed to revisit the implementation for more meaningful result. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: AGRAWAL Himalaya / COMP Course: UROP1100, Summer Dyslexia is a learning disorder which makes it difficult to read and interpret words and letters but does not affect general intelligence. In this project, our goal was to help pre-screen dyslexic students so that we could provide them with the extra attention and assistance they may require in order to read and learn effectively. This was to be done by using the data collected in a Chinese writing application developed for dyslexia prescreening, where the students would write the characters as instructed, in the hopes of finding any relationships between the data and dyslexia in particular students. 101 School of Engineering Department of Computer Science and Engineering AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: SZE Kai Tik / COGBM Course: UROP3100, Summer This paper reports the development of a Guided Wire Recognition System for Tseung Kwan O Hospital (TKOH). In this project, the challenges and concerning solutions in the training process of the YOLO v5 model will be highlighted, and the current development process will be updated, especially in the stage of user acceptance test, we understand more regarding the customer needs and re-customize our model training based on the hook recognition. In addition, the paper will raise some reflections on the current stage and put forward ideas about follow-up research. AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data Supervisor: CHAN Gary Shueng Han / CSE Student: PENG Zhuoxuan / COSC DAI Tianyuan / COSC Course: UROP4100, Summer UROP1100, Summer Crowd counting is to approximate the number of objects in an image or a video of unconstrained scenes. Current state-of-art method utilized in crowd counting is based on convolutional neural networks (CNNs). Despite satisfactory performance, unreality of exact counting of objects is still the fact due to large variation in object scales and isolated small clusters of objects. To preserve long-range context information when generating density maps, multi-scale architectures are embraced by several researchers. However, long training time, non-effective branch structure and scant of dots in density maps also bring about great challenges. Recently, some researchers argue using the Transformer architecture in vision tasks, mirroring its success in the field of nature language processing. Unfortunately, high resolution of images and largescale variation are obstacles waiting to be solved. In this paper, we compare currently prevailing models in crowd counting by setting our own pipeline and metrics on different datasets. Indoor Localization and Mobile Computing Supervisor: CHAN Gary Shueng Han / CSE Student: LIU Dingdong / COMP Course: UROP1100, Fall In UROP1100 - Indoor Localization and Mobile Computing project, I learned about the common practices in Indoor localization. I analyzed an existing project, which is the final year project done by other students. Based on the analysis of this previous final year project and another set of data collected by the Microsoft team, we discussed the feasibility of a new indoor localization approach, which uses map information to calibrate the sensor data. Finally, we concentrated on an existing indoor localization problem, namely the floor classifying problem, and my mentor provided a novel solution to it. In this report, I will introduce the work I have done and elaborate on my experiment's analysis. This project serves as an extension of my summer COMP4900 course. In this project, I continued to learn about crowdsourcing methods and have an experiment on them. Hopefully, in the continuing sections of UROP, I can dive into the cutting edge and raise my ideas. 102 School of Engineering Department of Computer Science and Engineering

Indoor Localization and Mobile Computing Supervisor: CHAN Gary Shueng Han / CSE Student: HU Chenxi / COGBM Course: UROP1100, Spring Indoor localization has always been a significant field for its commercial interests. As GPS signals are feeble inside buildings, other signals such as Wi-Fi are utilized to localize users. Among all the techniques, Wi-Fi fingerprinting has been the most promising one for its high applicability in complex indoor environments. Euclidean distance, cosine similarity, and Tanimoto similarity are widely adopted to compute the user’s location in the deterministic method. Aside from previous mathematical methods, Autoencoder also has good applications in similarity search with its dimension deduction. Hence, the research team explored the possibilities of adopting Autoencoder in the Wi-Fi Fingerprint-Based Indoor Localization field. The report aims at demonstrating the performance for Indoor Localization implemented with Autoencoder. Indoor Localization and Mobile Computing Supervisor: CHAN Gary Shueng Han / CSE Student: GUO Zhi / SENG Course: UROP1000, Summer During the three months in the summer term, I have been doing the UROP1000 project: Indoor Localization and Mobile Computing by professor CHAN Gary Shueng Han. In this research course, I have learnt a lot of things about doing research. I tried to read and understand papers on different topics for the first time. I found that many techniques used in the papers are totally new to me. So, I worked hard to learn this knowledge then reread those papers, trying to have a better understanding of them. Although I didn’t really accomplish anything useful in this project, the knowledge I got during this process still means a lot to me. Here are the details. Indoor Localization and Mobile Computing Supervisor: CHAN Gary Shueng Han / CSE Student: ZHANG Zhong / SENG Course: UROP1000, Summer This report analyzes and compares different methods that decode a rasterized format floor plan image. The work aims to sampling reference points on floor plans automatically. Methods including simple image processing algorithms and deep learning are discussed. Some of them are implemented and further tested using UST indoor map data. The report analyzed their advantages and disadvantages, discuss their usage, and potentially enlightening in indoor localization. In the end, I raised a method that integrated machine learning and simple image processing to do this work. It may help and bring some idea to the future development of automatically floor plan image processing. 103 School of Engineering Department of Computer Science and Engineering Video Analytics and RF People Sensing for Smart City and New Retail Supervisor: CHAN Gary Shueng Han / CSE Student: TANG Tianhao / COSC Course: UROP3100, Fall In this semester, I have learnt about graph neural networks (GNN) and its applications. I first got to know what a graph neural network is, what it is different and similar to convolutional neural networks, and how it works. Then I learnt how to build a graph neural network based on the framework. Furthermore, I have read some papers that focus on extensions of GNN like the heterogeneous graph. Besides, I have also participated in a project that is going to build a model to classify the spam users from the review data from 58, a famous site that provide services on lives. This report will state them in detail. Video Analytics and RF People Sensing for Smart City and New Retail Supervisor: CHAN Gary Shueng Han / CSE Student: TANG Tianhao / COSC Course: UROP4100, Spring In this semester, I continuously focus on dyslexia or dysgraphia project. This semester, we mainly focus on using computer vision and machine learning technologies to identify the faults and problems in writing characters, like missing components or flipped the character. To achieve this, we first tried building a system to distinguish the chirality of writing characters globally or partially. We then tried several ways to develop a system that can decompose the character components and strokes and giving out problems targeting each part of the characters. This report will give details on what we have done and what we will do in the future. Video Analytics and RF People Sensing for Smart City and New Retail Supervisor: CHAN Gary Shueng Han / CSE Student: CHANG Hong-yuan / DSCT Course: UROP1000, Summer Videos captured by cellphones or surveillance cameras often suffered from blurriness, which motivates research interests in video super-resolution (VSR). The primary goal of VSR is to computationally generate high-resolution videos given corresponding low-resolution inputs. In this UROP project, we specifically interest in large scaling factors VSR for real-world video clips. We first briefly discuss the problem setting of super-resolution (SR). Next, we introduce three state-of-the-art models for SR problems and conduct several experiments. Finally, we analyze the pros and cons of different models qualitatively and suggest possible directions for future improvement. 104 School of Engineering Department of Computer Science and Engineering Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: CHONG Cheuk Hei / COMP Course: UROP1100, Fall This research project has been involved in developing simple decentralized mobile applications built with Flutter, a cross-platform mobile frontend framework, to interact with smart contracts powered by Solidity programming language in the Ethereum test network, which simulates depositing and withdrawal function in DeFi applications. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: EDRIC Jovan / COMP Course: UROP1100, Fall Mobile Applications have long relied on conventional centralized servers for storing and processing data. As concerns over centralization have risen and the need for decentralization has increased, blockchain technology seems poised to replace the backend infrastructure of traditional centralized applications. The convenience of centralized servers, however, is still a major selling point for most developers. The proposed semi-decentralized application offers a platform in which users can offer freelance service to potential clients of the app. It adopts the Ethereum blockchain for securing transactions and user authentications, while employing an online centralized database, for storing additional data, including user and service information, benefiting from the convenience of a centralized server and security of a blockchain. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: MA Jianheng / RMBI Course: UROP1100, Fall One of the prevalent applications of using blockchains nowadays is smart contract. Solidity is an high-level and object-oriented language to implement smart contracts. It was developed based on C++, Python and JavaScript, and is designed to aim at the Ethereum Virtual Machine (EVM). To enable users to easily interact with the smart contracts, web3.js is used to developed frontend application. This reports uses both remix and Ethereum Studio to demonstrate several examples. 105 School of Engineering Department of Computer Science and Engineering Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: RUBAB Tamzid Morshed / DSCT Course: UROP1100, Fall This article is a progress report of my project on decentralized android application. The application is an online two player Tic-tac-toe game. Its back-end is implemented with Solidity programming language and front-end is under development with XML, Java, and Web3j in android studio. It supports creating and managing accounts for each player, creating a game and password protecting it, joining a game created by a friend or stranger, taking turn in the game, and saving the statistics after the game ends. The back-end supports any n×n board game, so the front-end can be extended to include more Tic-tac-toe type games. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: TSANG Lok Kan Ethan / QFIN Course: UROP1100, Fall This paper is writen based on my previous publication of ”BCred - A Blockchain Platform for Academic Credentials”. BCred is an blockchain application idea proposed on the International Blockchain Olympiad 2020, intended to develop a complete, efficient, yet secure system for academic credentials. In this paper, I will illustrate the basic set up and cautions for general decentralized mobile applications development. In Section 1, I will give a brief introduction on distributed ledger technology (DLT), blockchain, decentralized applications (DApp), in addition to BCred. In Section 2, I will frame the basic set up and prerequisite for decentralized application development. In Section 3, I will illustrate the simple implementation of decentralized application based on BCred. In Section 4, I will elucidate some cautions and guidline for decentralized application development based on two publicly trusted and recognised publications. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: EDRIC Jovan / COMP Course: UROP2100, Spring Blockchain technology was first introduced as a method of securing financial transactions. It has since evolved into something much broader. The advent of smart contracts, in particular, has brought into fruition the feasibility of decentralized applications, to run and execute turing-complete code in the blockchain, all while taking advantage of the tamper-proof nature of the blockchain. Despite this, the centralized nature of conventional internet applications offers more flexibility, as a tradeoff, when compared to decentralized applications. The proposed decentralised application offers a marketplace for consumers and service providers to buy and sell services, by adopting a middle-ground approach, securing transactions in the Ethereum blockchain and storing overhead user meta-data in a centralized cloud database. 106 School of Engineering Department of Computer Science and Engineering Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: RUBAB Tamzid Morshed / DSCT Course: UROP2100, Spring This article describes the remaining part of my project on decentralized android application. In the previous article (for UROP1100), I gave a documentation of the back-end, which is built on a smart contract and handles the creation and management of accounts and communications between players. Back then, the front-end only supported creating accounts and logging in. The current version also supports creating or joining a game, playing the game, and showing all statistics after a game ends in the front-end, which includes a graphical user interface, connection methods with the back-end, and the game logic. As the back-end supports any n×n turn based board games, the front-end can be updated to include more variations of Tic-Tac-Toe. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: BHATT Arun Datt / COMP Course: UROP1100, Summer The advent of distributed ledger technologies (DLT) has formed a basis for a decentralized mobile technology that, unlike conventional mobile applications, necessitates no central administrator. The practicality of the technology in its implementation and maintenance, and its dominance in security and anonymity has popularized the technology in banks, hospitals, and other sectors with a need for decentralized applicability. This report will explain my learning objectives, gains, and progress of building a decentralized mobile application on an Ethereum blockchain, one of the most popular example of distributed ledger technology. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: MA Tsz Fung / COMP Course: UROP1100, Summer Decentralized Application (DApp) is a computer or smartphone application that runs on Distributed Computing Systems. It works as an interface that allows users to access blockchains through Smart Contracts, a self-executing program that is located in blockchain (Smart Contracts and DApp, 2021). In this report, we will mention what blockchain is and how DApp is used in different online industries. We will also focus on Ethereum, one of the popular Smart Contracts platforms, to understand what is inside the Smart Contracts, and how the Smart Contracts do transactions. The report will also contain some codes of Smart Contracts programming language Solidity, and also the tools that can be used for Smart Contracts development. 107 School of Engineering Department of Computer Science and Engineering Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: PAN Hsuan-en / CPEG Course: UROP1100, Summer The goal of my project is to design an iOS app that interacts with a smart contract deployed on the ethereum blockchain. This mobile application allows the user to navigate on a world map to see descriptions of various famous tourist attractions, and then buy tickets to travel there. With the help of the characteristics of blockchain, the procedures are decentralized, automatic, and final. After successful payment (ticket price is in cryptocurrency: ether), the ownership of the ticket is assigned to the buyer, and could be easily looked up and confirmed afterwards. The benefits of this method are that it is highly convenient and the risk of people fabricating fake tickets is reduced to zero. The whole project can be divided into two parts - the mobile application and the smart contracts. Decentralized Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: WANG Yumeng / SENG Course: UROP1100, Summer Decentralized application is a new kind of application that is different from the usual applications we are using today. For the standard web applications we are using today, including Uber and twitter, are run by a central organization fully authorized, which means all of the customer taking its service will be in the control. For instance, some of the twitters sent by the individuals can be blocked and some of the account can be permanently suspended due to the decision of the company. This is also referred to as Web2. On the contrary, for the applications that are not centralized, often run on the block chain, no one is able to impose any restriction to anyone else. We can now build decentralized application on the block chain or a peer-to-peer (P2P) network. In the project we mainly focus on how the application is working with the android studio. To understand the mechanism of the decentralized application, we first start from traditional mobile app building skills base on android studio using the language Kotlin, and then we gain our understanding of ethereum block chain in an introductory manner. Furthermore, we develop the first decentralized app combining the knowledge. 108 School of Engineering Department of Computer Science and Engineering Introduction to Brain-Computer Interfaces Supervisor: CHATZOPOULOS Dimitrios / CSE Student: CHEUNG Sze Yuen / COMP Course: UROP1100, Fall A Brain-Computer Interface (BCI) is an interface which allows direct communication between brain and modern computers. Recent research has opened possibilities for BCIs focus on enhancing performance on healthy people. With the advancement of Deep Learning, it could largely alleviate the need of manually extraction of features from the signals. In this research we explored a generalized convolutional neural network for EEG signals processing on the dataset ‘Attempted arm and hand movements in persons with spinal cord injury’. Compared to the baseline, EEGNet performed similarly in terms of accuracies. The bp+LDA classification performed similarly also. Our result suggest that the rise of usage of Deep Learning could accelerate the exploration of EEG and BCIs. The code can be found at: https://github.com/Arcaneless/eegnet and https://github.com/kshatilov/eeg-aarhm-sci/blob/main/KEEGclf.py Introduction to Brain-Computer Interfaces Supervisor: CHATZOPOULOS Dimitrios / CSE Student: ZHU Chen / COMP Course: UROP1100, Spring Blockchain has become an extremely hot topic in recent years. When we study the architecture and functionality of blockchain, the smart contract is always an important topic. In this project, I studied the basic concept of blockchain, the principle of the popular blockchain Ethereum, building smart contracts using solidity and how we interact with smart contracts using web UI. My report will be a conclusion of what I have learned in this project. It will focus on two topics: the principle and advantages of blockchain and how to build and interact with smart contracts. Introduction to Brain-Computer Interfaces Supervisor: CHATZOPOULOS Dimitrios / CSE Student: LAI Kei Chit / MAEC Course: UROP1100, Summer Smart contracts are computer programs that are stored and executed on the blockchain. Although there are several limitations, smart contracts are increasingly popular in different areas because all the functions and states are stored on the blockchain without reliance on any intermediary. The decentralization, immutability, transparency and security differ them significantly from traditional computer programs. In this project, I have learned the characteristics and applications of smart contracts on the Ethereum blockchain through some readings. I have learned the Solidity programming language and have gained experience in developing and using some representative examples, such as bank, voting and auction, on Remix IDE. 109 School of Engineering Department of Computer Science and Engineering Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: FERDY / COMP Course: UROP1100, Fall Artificial Intelligence has been one of the biggest topics in society nowadays. The benefit of using Artificial Intelligence, such as machine learning, in everyday life is substantial and therefore all industries are in an arms race to incorporate AI to their products. The fast growth of AI startups around the world showing how companies try to maximize user experience by analyzing the input data entered by the user, record their usage pattern and user feedback. The proportion of people who use mobile phones also increases every year. Thus, the practicality of mobile development combined with sophisticated machine learning algorithms can offer users better experience and a wide range of innovative products. Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: TOH Magdalene Youjun / COMP Course: UROP1100, Fall Meal planning is important to reduce food wastage and ensure one has a balanced diet. However, one may run out of recipe ideas if one cooks frequently. While there is an abundance of recipes on the internet, it can be time-consuming to filter through them to discover recipes that are suitable. Hence, the purpose of this project is to produce a mobile application that gives personalized recipe recommendations to its users. The recommendation system utilizes content-based filtering to recommend recipes based on user preferences such as preferred preparation time, equipment, and types of ingredients. After candidate generation, the recipes will be further ranked according to the similarity of the ingredients in the recipes to the food supplies that the user already has. Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: WONG Yeuk Nam / IS Course: UROP1100, Fall Machine learning has been widely used in daily lives for more accurate predictions and higher working efficiency. With the developing technology on optical character recognition and the high bookkeeping cost among companies, this project aims at developing a mobile application that detects total price from receipts for automatic bookkeeping records. This project has selected Android Studio as its application development environment. To develop the application, different online resources and cloud services have been evaluated. Given the limitations with time and app development knowledge, the approach is to build additional features on existing app resources that align with the basic features including the use of smartphone camera. After recognizing numbers from receipts, the application should be able to insert records into a data table that facilitates further processing such as analysis of monthly expenditure. The application is still under development and not ready to work smoothly on mobile yet because of the incapability between the original application and the additional features. 110 School of Engineering Department of Computer Science and Engineering Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: ZHENG Hantao / DSCT Course: UROP1100, Fall Mobile applications assisted by machine learning(ML) algorithms can offer sophisticated functions by adapting to the individual usage patterns of the users. In this project, we developed smartphone applications based on flutter, an open-source UI software development kit developed by Google, and integrate the ML. function with the help of Firebase, Google’s cloud computing and development platform. The report mainly focuses on the exploration during the term and the approaches made to solve the proposed problems, as well as the resources read for reference. We started with the basics of Flutter including widgets, navigators and other concepts for fundamental GUI design in mobile platforms. Then the powerful plugins Flutter provides such as flutter camera, image picker are introduced to our implementation. We finally tried to combine our application with the various functions supported by Google Firebase, including the cloud database service and face detection, text recognition and other powerful functions using ML algorithms. Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: KAO Shiu-hong / DSCT Course: UROP1100, Spring With the rapid development of modern technology, people often need to deal with much more information nowadays. In this report, we aim to build an app helping people categorizing articles, such as news, faster. We firstly collected hundreds of sample news on the Internet, and then used a machine learning algorithm to do multi-class classification, dividing these samples into three categories - business, engineering, and science. Finally, we built an Android app on Android Studio, uploading articles onto this app, and used the criteria we got in the machine learning to achieve our target. However, we found that the error rates in the training set and the testing set differ a lot. This may be caused by two reasons, resource collection and characteristic ignorance. We hope these problems can be fixed by considering more possible issues which may happen in the algorithm in the future. Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: TOH Magdalene Youjun / COMP Course: UROP2100, Spring Recommendations shape lives, from entertainment in videos or movies to purchases on e-commerce websites. This application of machine learning saves time by offering pre-filtered choices to the relevant choices. Recommendations may be applied in a food recipe mobile application. Diverse food recipes are abundant online. However, for everyday cooking, users will use recipes similar in cuisine, techniques, or ingredients. This project uses natural language processing to recommend recipes with higher similarity in ingredients and equipment needed. The two stages of the recommendation are candidate generation and ranking. 111 School of Engineering Department of Computer Science and Engineering Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: THIEN Zhong Vei / SENG Course: UROP1100, Summer The idea I have in mind for the UROP project is to build an app for users to order different types of coffee using a mobile device. A pop-up message containing information about which type of coffee is most suited for the user is display as soon as the mobile application is run. Nevertheless, users can still choose coffee of their liking despite the suggested selection. In this context, the prediction of the best-suited coffee would be decided by the current time and weather. Machine learning on the other hand is being implemented to predict the best type of coffee based on all the previous selection of the users of the app. Whenever a user selects a coffee, a dataset will be created and saved to the database to improve the accuracy of the regression analysis. Practical ML-based Mobile Applications Supervisor: CHATZOPOULOS Dimitrios / CSE Student: ZHENG Tianshi / COGBM Course: UROP1100, Summer In this project, I aimed to build a task management app in React Native. The app collects task finished information and data from the user and generate a report based on it. It is a minimum viable product of a mature full-stack task-management app. It includes a front-end UI, backend model and API built by flask RESTful, and it also applies Artificial Intelligence techniques such as Natural Language Processing, to analyze the tasks by category, etc. Data-Efficient, Domain Generalizable and Interpretable Deep Learning Supervisor: CHEN Hao / CSE Student: HUANG Junkai / COSC Course: UROP1100, Summer This experiment is inspired by the Mitosis Domain Generalization (MIDOG) Challenge 2021. Mitotic figure density is an important criterion for tumor diagnosis and prognosis. With the advancement of deep learning, contemporary deep learning-based algorithms for mitotic figure identification can now match the performance of human specialists. However, there is a bottleneck: domain-shift impairs the performance of deep learning models. We undertake a comparison experiment to assess the degree of loss in mitotic figure identification ability using state-of-the-art deep learning models to better demonstrate this concern. In this experiment, we put YOLOv3, Faster R-CNN, and Cascade R-CNN to the test. Then examine the results of the tests. The codes are available at: https://drive.google.com/drive/folders/1zvDqtN6Zgx9UcYVYFSI-12_QjFLoozKG?usp=sharing 112 School of Engineering Department of Computer Science and Engineering Deep Learning in Medical Image Analysis Supervisor: CHEN Hao / CSE Student: LAU Ying Yee Ava / DSC Course: UROP1100, Summer Osteoarthritis is a very prevalent disease, which creates a huge demand for automated and accurate assessment of osteoarthritis status. Past studies has looked into creating a reliable system, by using different bioindicators as a benchmark. These joint space area, minimum joint space width, osteophyte area, tibiofemoral angle, cartilage measurements, bone texture and bone size. However, the scoring is not accurate enough to reflect the severity of osteoarthritis and be able to be applied in real life. There are 3 potential directions to improve the scoring system: a multiple-model system with MRI and CT scans, using multiple bio indicators, and by reviewing the scoring method. Deep Learning in Medical Image Analysis Supervisor: CHEN Hao / CSE Student: OW YONG Chee Seng / CPEG Course: UROP1100, Summer Due to current advancements in technology, the applications of computer vision had expanded to the domain of biomedical science, and it seeks to automate tasks that the human visual system does with higher precision and accuracy. In this research, Semantic Segmentation is applied to locate and highlight different layers of Optical Coherence Tomography scans (OCT). For this task, I utilized a fully convolutional network model for the task, which is U-NET architecture to localize areas of OCT images. The rationale behind using this architecture is due to the ability of U-NET to localize and distinguish layers through segmentation of every pixel in the input image and funnel them to their category of classes. In this case, 6 classes were used as the model segments the 6 most obvious layers in OCT scans. Through the use of Annotated Retinal OCT images database (AROI Database) from Sestre Milosrdnice University Hospital Center, I have a total of 1105 OCT scans with both raw and masked as the initial dataset. After training the U-NET model with 50 epochs, the accuracy for train data is 0.98 with loss of 0.03, while testing on new images shown that the dice coefficient is 0.9 and the mean intersection over union score is 0.87. This shows that it considerably performs well over untested data. However, it is important to note that the training set had very similar images and masks as they are all scanned from 24 patients, and resource limitation towards the deep learning model. With this in mind, I am interested to further develop segmentation and localization of OCT images such that abnormalities (like Diabetic Macular Edema) can be detected with high accuracy and treatment can be provided. 113 School of Engineering Department of Computer Science and Engineering Deep Learning for Ophthalmology Image Analysis Supervisor: CHEN Hao / CSE Student: CHEN Siyu / ISDN Course: UROP1000, Summer 2D fundus images and 3D optical coherence tomography (OCT) scanning volumes are two modalities of clinical data used for glaucoma grading. In this project, we use deep learning architecture to classify the samples of eye images into three categories according to visual features of these two modalities. The dual branch network was trained on data from 2D fundus images and 3D OCT images separately. The model can classify modality images into no glaucoma (without glaucoma are not samples without diseases, but images from patients with other eye diseases), early glaucoma, and moderate or advanced glaucoma. This classification methods can be an inspiration of eye disease diagnosis based on multi-modalities images using deep learning network. *The dataset and related materials used in this project are provided by ‘MICCAI2021 Contest: GAMMA’ which holds the copyright. Machine Learning in Healthcare Supervisor: CHEN Hao / CSE Student: KIM Hanbin / BIBU Course: UROP1000, Summer This report explores the application of machine learning in the healthcare industry, especially in drug discovery. Machine learning algorithms are being used in numerous aspects of drug discovery such as drug candidate selection and prioritization, and drug toxicity prediction. Through Computer-Aided Drug Discovery, pharmaceuticals could reduce R&D and clinical costs; Besides, an increasing trend of partnerships between Big Pharma and AI vendors could be seen. Among many aspects of drug discovery, this report investigates the use of machine learning in drug toxicity prediction by using Tox21 Challenge dataset and executing random forest model of a challenge winning team. This report also further gave a chance to think about the promising prospect of machine learning and its influence in drug discovery processes. Machine Learning in Healthcare Supervisor: CHEN Hao / CSE Student: SINGH Har Shwinder / PHYS Course: UROP1000, Summer Brain tumor detection is essential to identify brain cancer. This method aims to use semantic segmentation for AI tumor segmentation on MRI scans. A 3D Attention U-Net was used to train the 3D segmentation task of the BraTS 2021 Challenge Task 1. The Attention U-Net model improves the basic U-Net as the attention module increases focus on relevant feature maps, making the training more efficient. The model was trained with deep supervision, and data augmentation on the training dataset. Notably, positively labelled segmentation samples were cropped out as additional training samples. The highest results performance as submitted to the validation phase was: Dice of 0.817, 0.86 and 0.908 and Hausdorff 95% of 14.5, 9.9 and 5.97 for the enhancing tumor, tumor core and whole tumor respectively. 114 School of Engineering Department of Computer Science and Engineering Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: PARK Chun Ho / MATH-CS Course: UROP1100, Fall Although the problem of super-resolution has been explored thoroughly, super resolution of polarization images is an interesting problem that has not been addressed yet. Polarization images has numerous industrial applications as it allows tasks such as damage detection and reflection removal, therefore developing a super-resolution model on polarization images would allow improved performance and novel extensions on performing these tasks. This project aims to test existing super-resolution models developed for conventional images to polarization image dataset, and find out which models yield reasonable performance on this specific type of data. For this study, the SRCNN model and the EDSR model have been tested, and the EDSR model has shown superior performance in terms of PSNR and SSIM scores. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: REN Xuanchi / COSC Course: UROP1100, Fall We present a self-supervised approach with pose perceptual loss for automatic dance video generation. Our method can produce a realistic dance video that conforms to the beats and rhymes of given music. To achieve this, we firstly generate a human skeleton sequence from music and then apply the learned pose-to-appearance mapping to generate the final video. In the stage of generating skeleton sequences, we utilize two discriminators to capture different aspects of the sequence and propose a novel pose perceptual loss to produce natural dances. Besides, we also provide a new cross-modal evaluation to evaluate the dance quality, which is able to estimate the similarity between two modalities (music and dance). Finally, experimental qualitative and quantitative results demonstrate that our dance video synthesis approach produces realistic and diverse results. Source code and data are available at https://github.com/xrenaa/Music-Dance-Video-Synthesis. 115 School of Engineering Department of Computer Science and Engineering Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: SUN Yushi / COSC Course: UROP1100, Fall In the field of photography, it's always desirable to remove the polarized reflection from the image, since the reflection actually influence the visual quality of a photo. However, this is not an easy task, since it's hard for machine, even human, to tell which part belongs to the reflection that we want to remove and which part belongs to the original image. In this project, we want to identify the potential of employing super resolution models to perform polarized reflection removal task. We will first build up some existing super resolution models and then test them on the Polarized Reflection dataset provided by [1]. The model structure and the experiment result will be demonstrated. Since this is just a progress report, we will also introduce our future plan at the end of this report. [1] C. Lei, X. Huang, M. Zhang, Q. Yan, W. Sun, and Q. Chen, “Polarized reflection removal with perfect alignment in the wild,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 1750–1758. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: ZHOU Ji / CPEG Course: UROP1100, Fall The project focuses on applying super resolution models on multiple data sets. With the testing and training of data sets based on polarized reflection removal and multiscopic vision, we expect to improve the original model and increase the accuracy of the model. In this project, I intended to use SRGAN as the SR model. Unfortunately, due to the lack of knowledge about machine learning and computer vision, and the tight time schedule, the final outcomes is not as well as we expected. This report will introduce related resources such as the applied model and the data sets used with a short explanation and show the process of the project in this term. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: JIANG Xudong / COSC Course: UROP1100, Spring With the development of deep learning, various computer vision tasks have made significant progress. One of the important computer vision tasks, image and video superresolution, also experienced remarkable improvement. Various deep learning models have been proposed to solve the video superresolution problem. However, there is a popular video format, GIF, which most existing models cannot perform well. This result from the artifacts caused by severe color quantization and color dithering. In this project we review recent deep learning techniques about video super resolution and extend the application to GIF format videos. We obtained some perceptually more satisfying results in the experiments, although the result in objective evaluation metrics is still in need of improvement. 116 School of Engineering Department of Computer Science and Engineering Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: LI Kwan To / COMP Course: UROP1100, Spring We present a simple deep learning model for image background removal. By incorporating a hint background image, we are able to get reasonable good performance for the segmentation task. The technique could be useful for extracting foreground person implemented in virtual background functionality in conferencing software. We adopted a variant of the famous U-Net which is widely used for segmentation tasks in the literature. Experiments are done to carefully tune the hyperparameters and explore regularization techniques to improve accuracy. Our approach strikes a balance between low-quality simple segmentation and high-quality background matting. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: ZHOU Ji / CPEG Course: UROP2100, Spring In this term the project focuses on deep learning methods for high dynamic range (HDR) image and video synthesis. HDR is a new technology used to capture and reproduce scenes with high dynamic range. By applying deep learning methods, we might be able to reconstruct HDR images or videos without HDR raw materials. To achieve this target, we start from multi-view HDR image synthesis, then to single image reconstruction and finally go to HDR video synthesis, based on the sequence I learned in this semester. This report will introduce basic concepts of deep HDR imaging of dynamic scenes and HDR image reconstruction from a single exposure, which come from image synthesis theories. Then we move to HDR video reconstruction and our process dealing with real world benchmark datasets. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: WU Chi-hsuan / DSCT Course: UROP1100, Summer Low-resolution (LR) textual images are often generated after file compression. Recognizing these blurry texts in LR images are challenging especially when the neighboring texts could not provide enough information. In order to recover a high-resolution (HR) image given a LR image, super-resolution (SR) techniques were introduced. Although SR techniques have been developed for years, research in SR techniques on documentary textual images are still insufficient. In this report, we created a documentary textual image dataset and test the SR effect on these images with a scene text recognition (STR) model. Our goal is to observe the effectiveness of SR on documentary textual images using scene text SR model, and provide a reference for future textual images SR studies. 117 School of Engineering Department of Computer Science and Engineering Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: WU Yi / COSC Course: UROP1100, Summer Text image super resolution is an important sub-field of Super Resolution, which aims to recover the realistic features of a text image from a low-resolution text image. Recovering the realistic features from a low-resolution image is challenging and it has drawn much research interest of the computer vision community. In our project, we self-learned some basic knowledge of deep learning from cs 231n and collected thousands of screenshots as a dataset, then use this dataset to train two open-source deep learning models. The training result shows these models show surprising performance on highly blurred screenshot images, and we plan to improve these models to achieve higher performance. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: XIAO Ziruo / SENG Course: UROP1100, Summer In the scene of text image got zoomed or compressed, low-resolution often results in the loss of text information. Our group focus on the super-resolution of text image, trying to get the text clear enough to be recognized. We collected data, run them on some existed models, trying to figure out their shortcomings and hopefully build one better for Chinese text in the future. As a year 1 student introduced to computer vision and deep learning for the first time, I put most of my effort in learning and practicing the basic knowledge of the convolutional neural network in this summer. This report will show what I have learned in this 3 months, the project progress and my future plan. Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: YAO Chongchong / DSCT Course: UROP1100, Summer As compression technology is commonly used, compressed text pictures may be blurred and difficult to read. In such cases, it would be helpful to restore their original sharpness, which is where super-resolution technology comes in. This project aims to learn and apply super-resolution models to computer-generated text images to increase their resolution. To achieve our goal, our team managed to collect naïve data and processed them to generate training and testing samples. We first tested our data on several pre-trained models and compared the results. In the end, we successfully trained several models and got satisfying sharpened outputs. Overall, this study produced some practical results for computer-generated text image super-resolution and also improved our knowledge and technique on deep learning. 118 School of Engineering Department of Computer Science and Engineering Deep Video Super-Resolution Supervisor: CHEN Qifeng / CSE Student: ZHANG Juntao / SENG Course: UROP1100, Summer Text recognition is one of the earliest research fields in computer vision, aiming to enable computers to recognize text in various scenarios. As image information explodes on the internet, recovering text information from blurred pictures and providing readings for real-world automated machines become needed. Text super-resolution comes into rising as a vital branch. This report first summarizes the current practice, including related neural networks like CNN and RNN, image super-resolution method, and a scene text recognition method by Chen et al. (2021) Then give a research report focusing on text super-resolution on how to blur pictures in training can effectively train the super-resolution model. Lossless Point Cloud Compression Supervisor: CHENG Siu Wing / CSE Student: REN Zhengtong / COMP Course: UROP1100, Fall An effective, efficient, and lossless compression algorithm can facilitate the storage and transmission of large point clouds generated by modern 3D imaging devices. Current compression algorithms are based on octree, binary tree, clustering, etc. However, so far there is no attempt at using Steiner tree, a generalized version of minimum spanning tree, to solve this problem. In this UROP project, we first introduce a custom encoding method for floating-point numbers. Based on this encoding method, we then construct a rectilinear Steiner tree for the point cloud using the 1-Steiner heuristic. This progress report will describe the framework of our algorithm, highlight experimental results collected so far, and point out directions for future work. Automated Program Synthesis Supervisor: CHEUNG Shing Chi / CSE Student: CHEN Yijia / COSC Course: UROP1100, Spring Currently reporting HKUST Undergraduate Research Opportunity Program project on Automated Program Synthesis mainly focuses on polishing and extending an existing idea presented in the paper 'Programming by Example made Easy' to be submitted. Programming by Example (PBE) is one of the popular approaches to the long-lasting research problem of automated program synthesis, and with its potential further developed by the novel bi-directional search synthesis algorithm implemented in the PBE framework 'BEE' proposed by the paper, this UROP project studies some general knowledge about automated program synthesis, further compared the performance of the 'BEE' framework to some other methods, and partially extended the usage scenario of the 'BEE' framework to an unconventional area: Minecraft gaming. 119 School of Engineering Department of Computer Science and Engineering Automated Program Synthesis Supervisor: CHEUNG Shing Chi / CSE Student: LUK Sui Hei / COMP Course: UROP1100, Spring In a computer system, coding is a convenient way to manage and manipulate bulk data, which would be tedious, error prone and time consuming to be done manually. However, clerks who need to handle the bulk data can have no knowledge nor experience in coding. With the advancement in AI and software technologies, it would be possible to synthesise programs without writing code through programming-by-example (PBE). In PBE, a program is synthesised based on a few examples provided by the users demonstrating their intentions. It is necessary to provide users a graphical user interface for interacting with the PBE system. The rationale is users could be illiterate in coding and it would be difficult for them to access through a command-line interface. Therefore, to facilitate a ease in employing the system, this project aims at developing a frontend graphical user interface for the system. In the field, there are similar project by Microsoft. Automated Program Synthesis Supervisor: CHEUNG Shing Chi / CSE Student: LU Junxian / DSCT Course: UROP1000, Summer In this session of UROP, I participated in the Automated Program Synthesis project supervised by Prof. CHEUNG. For most of the time, I work with Mr. WU Jiarong, who is a Ph.D. student working on this research. Soon after finishing exams of the regular term, I communicated with Mr. WU to better understand the actual target of the project and what background knowledge is needed. After the meeting, I understand that the goal of this project is to synthesize program analysis scripts. My main duty was to collect and prepare data. My first task is to collect real-world use cases of our Synthesis program in conference papers. In some of the related papers, the author provided code and data. So my second task is to check and make sure these artifacts (code and data) work. There are also some papers that don’t have artifacts available but described their procedure. My third task is to recreate the artifact according to the paper. In conclusion, I learn a lot about code analysis and related tools through different tasks I did in this UROP session. 120 School of Engineering Department of Computer Science and Engineering Building a Blockchain and Smart Contract Application Supervisor: CHEUNG Shing Chi / CSE Student: YANG Wenting / COGBM Course: UROP1100, Fall Ethereum can be regarded as a state transition model. Each smart contract has a state and the state change is triggered by transaction. To manage the smart contract, upgradability is an important characteristic. For developers, this characteristic can help them fix bugs and avoid vulnerabilities. Besides, it can also help developers to add new functionalities. For software engineering, the revision history can help them identify bug pattern, guide bug location and generate bug patches automatically. However, smart contract is not upgradeable because it cannot be changed after being deployed to the Ethereum network. Therefore, we usually use the proxy to help upgrade the version of smart contract in reality. In this project, we try to figure out the proxy smart contract on the Ethereum given specific Application Binary Interface (ABI). As blockchain is completely anonymous and very few contracts have associated GitHub, we can find the contract address if we find the smart contract proxy. After that, we can also retrieve the history or evolution of smart contracts. Building a Blockchain and Smart Contract Application Supervisor: CHEUNG Shing Chi / CSE Student: CHRISTANTO Nicholas / COGBM Course: UROP1000, Summer Smart Contracts are programs deployed in the blockchain that can be executed when certain predetermined conditions are met and verified. It is used to automate an agreement that results in an immediate expected outcome without intermediaries. However, once deployed on a blockchain, the address and the code of Smart Contracts cannot be modified. One solution is to re-architecture the code into separates contracts, Proxy Contracts. This research is about identifying Proxy Contracts and mining the revisions of the logic contracts behind the Proxy Contracts under the guidance of Wuqi Zhang. This report will firstly establish the background of the research, identifying the differences among conventional programs and Smart Contracts as well as the purpose of the invention of Proxy Contracts. During the research, Mr. Zhang has assigned some readings that are needed to achieve an understanding of the research topic which are written under the second section. Application of knowledge obtained during the making of this project to obtain historical implementations is also presented in implementations. In the end, reflections upon the project results and my research journey experience are elaborated. 121 School of Engineering Department of Computer Science and Engineering Building a Blockchain and Smart Contract Application Supervisor: CHEUNG Shing Chi / CSE Student: YU Cheuk Hei / COMP Course: UROP1000, Summer As Ethereum introduces its new blockchain together with a Turing-complete programming language, users can write smart contracts and deploy decentralized applications freely. Smart contracts are a type of Ethereum account that run as automated programs on the blockchain. Due to blockchains’ immutability, no amendments can be made to fix security flaws in smart contracts after deployment. A simple vulnerability may sometimes result in large amount of stolen funds. This project identifies one of the many vulnerabilities– Displacement attacks in front-running by sampling past transactions. This report provides explanation on how the vulnerability was located and how analysis was conducted. It also contains recommendations on how smart contract engineers can avoid similar mistakes in the future. A De-Polarization System for Social Media Supervisor: HUI Pan / CSE Student: YU Yue / QSA TILNEY-BASSETT Oktarian George / COMP Course: UROP1100, Fall UROP1100, Fall As a result of the continued growth of social media, the political landscape of many countries has become deeply polarised. At the heart of this division in beliefs lies the media corporations that play the important role of relaying information, but that increasingly cater to specific audiences that share similar political stances as those communicated by the network. Hence, this polarisation of opinions has led to the building of echo chambers where users are less receptive to opposing opinions and stop engaging in active critical thinking of the information being presented to them. The focus of this project is to build a tool, Debubble, that can increase a user’s awareness of media bias and let users reach opinions that are of opposing viewpoints. This is achieved by collecting data on a user’s reading patterns to identify potential echo chambers, and then lead users to information otherwise unreachable. 122 School of Engineering Department of Computer Science and Engineering A De-Polarization System for Social Media Supervisor: HUI Pan / CSE Student: YU Yue / QSA Course: UROP2100, Spring The report consists of two parts. The first project is the continued project from the last semester which aims to build a Chrome extension, Debubble, to help the users explore more ideas based on which they are reading on the news sites. We made some improvements including the UI reconstitution, A/B testing set, and user survey module on the basis of the last semester’s work and published the fourth version on the Chrome Store. The second project investigates the user activities in a Reddit community called WallStreetBet (WSB) in the timespan when the short squeeze of the stocks such as GameStop (GME), AMC Entertainment (AMC), and BlackBerry (BB) appears. By dividing the users into different groups based on the time they joined in the discussion in the WSB community, we measure the correlation between different user group’s activity and the stock ticker’s volatility. The research result reports the significant difference between the new users and the old users in terms of the user activities including the frequency of mentioning the stock ticker’s name and the frequency of using emoji. A De-Polarization System for Social Media Supervisor: HUI Pan / CSE Student: ZAU Ka Ming / CPE XU Xiao / SENG LIU Yuehuai / SENG Course: UROP1100, Summer UROP1100, Summer UROP1100, Summer The project we did is focusing on the online psychologist. Several accounts were chosen from Instagram. Then, we did some analysis between the users and the online professional to establish the relationship between the online material and the reaction of users. Firstly, we did a qualitative analysis for 30 accounts that therapists own. Then, we develop the codebook. After that, we rated many accounts to judge whether these accounts are more about emotional or informational support. Algorithms and Games in Android Devices Supervisor: HUI Pan / CSE Student: IP Si Hou / COMP Course: UROP2100, Fall With the rise of feed forward neural networks, a massive number of researches in computer vision immerses. Generative adversarial networks are one of the innovative approaches in recent years. By adapting the concept, the style of the image can be converted from one style to another by the usage of a discriminator and generator. CartoonGAN is one of the approaches to converting images from the real world into a cartoon style. In this project, we want to further investigate the style transfer of cartoons. We would like to perform a real-time style transfer of cartoons. We would like to make a real-time GAN that shows both temporal consistency and preserving the style and content of the input scene. 123 School of Engineering Department of Computer Science and Engineering Algorithms and Games in Android Devices Supervisor: HUI Pan / CSE Student: ZHAO Jiachen / COSC Course: UROP1100, Spring Gamification is a method to motivate users by applying game elements to non-game situations. Building a satisfying gamified platform has advantages compared to ordinary ones, for example, more user activities and better outcomes. However, an unsuccessful gamification may have countereffects, and waste the efforts paid on it. In this study, we conduct a qualitative research to do sentiment analysis on the posts from Meta Stack Overflow. We will use a pre-trained natural language processing (NLP) model Bidirectional Encoder Representations from Transformers (BERT) to analyze the post texts to see whether gamification of Stack Overflow is satisfying and find out a pattern between questions and answers. Algorithms and Games in Android Devices Supervisor: HUI Pan / CSE Student: ZHU Zhengjie / COMP Course: UROP1100, Summer It is well known that whenever people download a new application, the onboarding screen (show the key features of this application) is always the first part. Whether users continue to use the app is often affected by this small part. In other words, if the guide interface is good, it is likely to make user continue to explore this app. I collected 51 apps from the Google Store. A total of 417 UI elements from these guide interface. A questionnaire based on four criteria (later will be talked) was designed, 21 respondents were invited to complete the questionnaire, and 1,071 pieces of feedback data were collected. Then analyze the data to draw some valuable conclusions. Finally, a flow chart is designed for the developer's reference. Human Faces Generation using GAN Supervisor: HUI Pan / CSE Student: KHAN Umaash Ahmed / RMBI Course: UROP1100, Fall Although much work has been done in computer vision privacy protection field, there are not enough resources comparing the advantages and disadvantages of different methods in the privacy protection field. In this UROP project, we identify 3 privacy protection methods in computer vision applications and evaluate each one of them. The 3 methods identified are face cartoonization, facial attribute cartoonization (nose, eyes, lips) and FakeGAN. For evaluation, we focus on 2 main aspects. These are user experience and identifiable possibilities. For user experience, the output pictures should maintain as much original emotion and meaning as possible while for the identifiable possibilities, the output pictures should be as different from the original as possible. 124 School of Engineering Department of Computer Science and Engineering Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: ZHAO Yizhe / COSC Course: UROP1100, Fall Human-in-the-loop means that humans train, test, or tune the AI system to help it achieve more reliable results. Human-in-the-loop can be performed to teach the system what features to look for and guide the system to get more accurate answers. To introduce a method to interact with Machine Learning Model, we learned principles of Reinforcement Learning and Imitation Learning. In this project, we did experiments on DQN and Behavior Cloning in OpenAI Gym environment to better understand the strategies and parameters of these methods. We read the paper of InfoGAIL to capture the basic idea of this model and get a preliminary understanding of Human-in-the-loop in Reinforcement Learning. Next step of this project is to do more experiments on different models and read more cutting-edge papers to further explore this area. Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: ZHOU Xinrui / COGBM Course: UROP1100, Fall This project aims at introducing a method to interact with Machine Learning Model in a Web/Virtual Reality environment. The main tasks for this semester are to study GAN and algorithms for its different variations, also we reviewed and reproduced the model of some related papers that committed to this field. The research assistant Mr. Arthur Yau led weekly meetings where students presented what they have learned and implemented for the week. The main task for students were studying the materials provided by the research assistant and also implemented and reproduced some paper results and do some literature reviews. Each person in the group has a different level of understanding of the technology involved, and the research assistant distributed different tasks to us according to each person's actual situation and progress. In addition, the research assistant grasped a good schedule and arrangement, and used Asana as a collaboration tool so that we could all keep up with the progress. Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: TAI Sung Chit / COSC Course: UROP1100, Spring Data poisoning attacks pollute the training data of a machine learning model, impeding model accuracy and producing biased results. Label flipping and outliers are examples of such attacks. In this study, we attempted to remove outliers and flip mislabelled data by utilizing human feedback in the training process of a pretrained model after identifying the unclean data using an outlier metric. Our results show that this new method sees slight improvement in the accuracy of the model. This method of training may see use in crowdsourcing projects where data is easily mislabelled and may be unclean for model training purposes. 125 School of Engineering Department of Computer Science and Engineering Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: TSUI Yuk Hang / MATH Course: UROP1100, Spring The Deep Convolutional Generative Adversarial Network (DCGAN) model has a huge contribution in generating high degree realism images with large latent dimensions, which is not favorable for end-user to control and generate the images they want without the understanding of the meaning of each dimension. We combine the conditional DCGAN, Variational Auto-Encoder (VAE), and Principal components analysis (PCA) to create a User-Interface (UI) with low dimensions for generating realistic images. The research study aims to provide a method for generating high dimensional images through a low dimensional input to benefit end-user control and study how the deep learning network interprets the image generation process. Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: ZHAO Yizhe / COSC Course: UROP2100, Spring Human-in-the-loop means that humans train, test, or tune the AI system to help it achieve more reliable results. Human-in-the-loop can be performed to teach the system what features to look for and guide the system to get more accurate answers. In this project, we learn the concept of Offline Reinforcement Learning and Class Activation Maps. We implement a behavior cloning agent on Atari Breakout game and train it via class, then use CAM to visualize the agent understanding. We are using interpretable ML technique to show the process of learning and try to provide the visualization during learning then game AI/Robot trainer can adjust the model according to the learning progress better instead of only relying on the loss function. Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: ZHOU Xinrui / COGBM Course: UROP2100, Spring This project aims at exploring explainable machine learning and different interpret methods. The main tasks for this semester are to study related XAI topic and interpret methods for machine learning models and their different variations, then used Captum package to do interpretation on CNN and BERT model for text classification based on word importance and corresponding sentiment. Also, we reviewed many some related papers that committed to this field. The research assistant Mr. Arthur Yau led weekly meetings where students presented what they have learned and implemented for the week. The main task for students were studying the materials provided by the research assistant and also implemented and reproduced some paper results and do some literature reviews. 126 School of Engineering Department of Computer Science and Engineering Human in the Loop Machine Learning Supervisor: HUI Pan / CSE Student: TSUI Yuk Hang / MATH-CS Course: UROP2100, Summer This research study focuses on inheriting specific facial features from target image to reference image. In recent year, concern about face security is increased and many models such as StarGAN and StyleGAN provide solutions on inheriting the facial features. This research aims to propose a new model which may combine the characteristics of the two models mentioned above and provide a higher privacy for publics. Human-Drone Interaction Supervisor: HUI Pan / CSE Student: HUANG Yiming / COGBM Course: UROP1100, Fall Currently, there exist several open source simulators for drone which can be used for experiments. For example, Airsim, Paparazzi UAV, ArduPilot, Dronecode, LibrePilot, etc. Since the Airsim which is developed by Microsoft team, can be easily installed and utilized, we choose to conduct our drone interaction experiments based on it. However, its APIs are still not totally suitable for the conduction of the drone interaction experiments. To further improve interaction methods on the Airsim platform and set up a more realistic simulation environment for the drone control, this semester we focus on the control test based on Airsim and the comparison with the Tello drone, seeking the direction for future experiments. Human-Drone Interaction Supervisor: HUI Pan / CSE Student: LI Wenbo / COMP Course: UROP1100, Fall Augmented Reality (AR) is the technology that combines real and virtual objects and enables the users to interact with the objects in real time, which has been a popular research field with the development of different tracking and mapping technologies. Hardware of AR display has also been built in many smartphones, which brought it to more consumers and greatly increased its social acceptance. This project is intended to have a deeper understanding of AR with the motivation to design interesting AR-based Apps. In this report, I will analyze several projects of designing and evaluating AR interaction techniques, followed by the note I made in the progress of learning Unity and C# programming to obtain basic knowledge for game developing, as well as discussing the trends in AR. 127 School of Engineering Department of Computer Science and Engineering Human-Drone Interaction Supervisor: HUI Pan / CSE Student: HUANG Yiming / COGBM Course: UROP2100, Spring Nowadays the drone has been applied in multiple scenarios of society, especially the part which might need to face with some fatal dangers that may cause the loss of human’s life. However, the controllability of the drone is still an existing challenge that needs to be improved demandingly, in which the evaluation of the drone control system is the most difficult part which may consume the drone and other control devices. Therefore, a solution of building a virtual testing environment comes, which is based on the existing computer software like unreal and unity, by simulating the physical condition and the complex flying environment inside the simulator and collecting the data such as finishing time of each flying task and the collision information of the flying process, analyzing the collected data and then updated the existing control system and algorithm of the drone. Human-Drone Interaction Supervisor: HUI Pan / CSE Student: LI Wenbo / COMP Course: UROP2100, Spring Augmented Reality (AR) is a popular human-computer interaction technology that merges virtual objects into the real world. As the motion sensors built into mobile phones, this technology has become increasingly accessible to the users. This also enables Web pages to provide immersive Augmented Reality Experiences, which is called WebAR. This project will dive into the concepts, design and development of XR, especially AR, via notes made during the learning process, followed by concluding the work I did for paper crawling to construct an AR survey, as well as reporting the progress of the WebAR project I am doing based on the materials learned and proposing further work to be done with the project. Machine Learning on Wearable Devices Supervisor: HUI Pan / CSE Student: CAI Shi Zhan / DSCT Course: UROP1100, Fall In every year, there are many impressive papers published on ACM proceeding. We might want to know what the most popular trend in each year is, so this project intended to crawl the papers from the Association for Computing Machinery (ACM) library and then analyse them. In this project, we mainly focused on the symposiums on User Interface Software and Technology (UIST) from 2011 to 2020. 128 School of Engineering Department of Computer Science and Engineering Machine Learning on Wearable Devices Supervisor: HUI Pan / CSE Student: WU Chi-hsuan / DSCT Course: UROP1100, Fall As wearable devices become more and more handy and gradually increase its popularity, information around us is likely to be recorded and uploaded to the internet more easily in the future. At the same time, people’s consciousness of protecting their own privacy have raised in these years. In this report, we propose a method to protect people’s privacy by cartoonizing the faces on the image and modify specific facial segments. The cartoonization of faces provide deidentification protections of individuals in the image while preserving the user experience and satisfaction when sharing the protected images in online services such as social networks. First, we use face detection model and cartoonization model to get the cartoonized faces. Next, we get the contour of different segments of the faces and crop them into ready-to-use images. We believe distorting facial segments could further protect people’s privacy on the image. Machine Learning on Wearable Devices Supervisor: HUI Pan / CSE Student: DING Ziheng / CPEG WANG Ziyi / SSCI Course: UROP1100, Summer UROP1100, Summer The project ‘Machine Learning on Wearable Devices’ is aiming at exploring interaction between users and programs on wearable displays through hand gesture recognition. It consists of 2 parts: ‘Copy and Paste’ and ‘Hand Pose Estimation by IMUs’. ‘Copy and Paste’ project is mainly about recognising hand gestures by camera to do ‘copy’ or ‘paste’ tasks, without any contact device. ‘Hand Pose Estimation by IMUs’ is mainly about recognising hand gestures by Inertial measurement units(IMU), which serves as an addition when hands are out of the detection range of the camera. Part I is made by WANG Ziyi and Part II is made by DING Ziheng. Parameterized Algorithms for Static Program Analysis Supervisor: KAFSHDAR GOHARSHADY Amir / CSE Student: BIDAIBEK Sanzhar / COMP Course: UROP1100, Spring During this UROP1100 course under the guidance of Professor Amir Goharshady, I successfully studied the introduction into parametrized algorithms as applied in the context of static program analysis, as well as the basics of infinite automata theory, and was assigned and started the work on a research topic of optimizing methods related to HyperLTL Model Checking. I conclude that this was a fruitful experience and I am eager to continue the work in the upcoming semesters. 129 School of Engineering Department of Computer Science and Engineering Parameterized Algorithms for Static Program Analysis Supervisor: KAFSHDAR GOHARSHADY Amir / CSE Student: CHENG Tien-chun / COMP Course: UROP1100, Spring Arranging frequently accessed data closely in the memory has long been considered an important program optimization. The relationship of data accesses can be model as a set of hierarchical clusters. In this article, we research an efficient algorithm to compute the data locality hierarchy parameterized by treewidth. We analyze a specific pattern of tree decomposition and perform a polynomial reduction on the parameterized problem to an NPComplete problem. We obtain a computational hardness result of measuring the closeness of two data under this hierarchical model. This result can be a major step toward proving the computational hardness of determining the closeness of a data group. Parameterized Algorithms for Static Program Analysis Supervisor: KAFSHDAR GOHARSHADY Amir / CSE Student: LAM Chun Kit / COMP Course: UROP1100, Spring In this project, we studied problems in static program analysis, trying to find useful parameterization for the problems and speedup existing algorithms. This report focuses on linearizability checking with program traces using strong d-hitting families. We proved the computation of the heuristics would take exponential time in the worse-case and devised an algorithm with slightly better upper bound on the number of generated schedules. Parameterized Algorithms for Static Program Analysis Supervisor: KAFSHDAR GOHARSHADY Amir / CSE Student: WU Zhiang / COSC Course: UROP1100, Spring Static analysis focuses on proving specific properties of a program via the argument on its code directly. Lots of problems in this field can be solved by parameterized algorithms with a better efficiency than traditional methods. In this semester, we first did a literature review on parameterized algorithm, automata on infinite words and basic model checking, then picked the numerical domain analysis which is a classical problem in static analysis to try to design a parameterized algorithm to solve it. In the end, some primal ideas were raised and we will try to solve the problem in the future. 130 School of Engineering Department of Computer Science and Engineering Parameterized Algorithms for Static Program Analysis Supervisor: KAFSHDAR GOHARSHADY Amir / CSE Student: WU Zhiang / COSC Course: UROP2100, Summer Treewidth is a popular parameter to tackle some practical NP-hard problems. Optimal cache management is one classical problem in the area of programming language and verification. Parameterized complexity is dramatically helpful for those classical hard problems, which always gives a way to solve the problem with better time complexity. In recent topics such as optimal mining for Bitcoin miners, the method of tree decomposition also improves the algorithm efficiency. In this UROP, we first have a rough understanding of how to use tree decomposition and then try to come up with an algorithm to compute reference affinity by designing a nice tree decomposition. Data Mining in Bitcoin Transaction Data Supervisor: LUO Qiong / CSE Student: KOON Tsz Hin / QFIN Course: UROP1100, Summer The Bitcoin network experienced a rapid growth since its introduction by Satoshi Nakamoto in 2008. In recent months, the sky-rocketing adoption and price increment of the Bitcoin attracted wide attention. In this study we extract blockchain data through running a Bitcoin full node and attempt to provide a general picture of the network. We then discuss our and highlighted our key findings in Difficulty, which is found to be an algorithm-based parameter that stabilizes the block generating rate of the decentralized network; SegWit, which introduces us the new concept of block weight and found to be scaling up the network and Patoshi Pattern, which demonstrates an unusual mining pattern in the early blocks. Analysis and Understanding of User Behaviors in Online Communities Supervisor: MA Xiaojuan / CSE Student: MOK Nga Yiu / SENG Course: UROP1100, Summer During the COVID-19 pandemic in the early 2020, there were numerous patients decided to seek help online, especially on Sina Weibo. This study classified the types of posts into four genres, while identifying the channels help-seekers utilized when posting. Thereby suggesting most of the patients intended to post their help-seeking posts on the help seeking super topic, however, some users may fail to do so, resulting misused help seeking super topics. This study also evaluates the differences between different channels, and hence suggested that help seeking super topic might be the best channel during the severe pandemic after evaluating different channels. 131 School of Engineering Department of Computer Science and Engineering Analysis and Understanding of User Behaviors in Online Communities Supervisor: MA Xiaojuan / CSE Student: WU Zhiyang / COSC Course: UROP1100, Summer In 2020, the outbreak of COVID-19 in Wuhan, China caused a severe shortage of medical resources at the early stage. As a result, hospitals were struggling to cope with the volume of patients. Many COVID-19 patients (including suspected patients) who were not accepted or treated by hospitals, turned to social media sites for seeking assistance. In this project, we collect help-seeking posts from different online communities in mainland China, including Weibo, Toutiao, and some self-created websites, and analyse help-seeking behaviors during the outbreak of COVID-19 pandemic in Wuhan, China at the beginning of 2020, with a focus on the information collected from Weibo. Exploring College Students’ Self-identity via Social Media Supervisor: MA Xiaojuan / CSE Student: XU Mingshi / COMP Course: UROP1100, Spring COVID-19 has influenced our lives fundamentally. It also profoundly reconstructed the behavior of Chinese university students, as college students usually use social media, especially Weibo, to update their activities and express their feelings and opinions in real-time. Therefore, these contents reflect their daily life and their mental state. In addition, some researchers found that activities on social media networks (i.e. tagging and sharing) provide insights into the self-identity of users. Therefore, we start using social media content to study the impact of COVID-19 on college students 1 and compare the reaction between students in different identity groups. Our understanding of the impact of the pandemic on users’ social media engagement will facilitate identifying users vulnerable to crises and the subsequent provision of appropriate psychological support by social media. Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: CHEN Zixin / DSCT Course: UROP1100, Spring Users today have instant access online to large amounts of information on which to base their decisions. For example, they can choose products or book hotels based on user reviews or system recommendations. However, such information can be conflicting, biased, subjective and scattered across many sources. Solely relying on automatic approaches to address these issues can be insufficient due to inadequate usage of context and constantly updated users needs or criteria. This project aims to explore novel, transparent designs that increase user satisfaction, engagement and efficiency with actual decisions. We built interactive prototype systems and evaluate their effects in users decision-making process. Personally, I participated in the second half of the whole project and mainly responsible for the data collection, labeling and processes, web page & database construction, mock experiments and part of final results analysis. 132 School of Engineering Department of Computer Science and Engineering Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: NAN Xi / DSCT Course: UROP1100, Spring Online platforms such as Reddit's 'Subreddit' and Weibo's 'Super topic' serve as digital fandom communities where users can seek information and discuss their idols. The rapid development of social media increases the possibilities of potential members joining the fandom community. Celebrities can suddenly attract attention from the public because of certain events. For example, Psy, a South Korean singer, was known domestically for his humorous videos and stage performances, and internationally for his hit single 'Gangnam Style'. In this case, a sudden influx of newcomers joins the fandom community and these newcomers possibly influence the community culture and interaction patterns. In this project, we try to characterize the user behavior in online fandom communities during the event-induced period. Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: YIN Zhuohao / DSCT Course: UROP1100, Spring With various social media platforms developing at a rapid pace, people are able to access much more information and thus more likely to participate in discussions in serendipitous groups, especially when a viral event concerning certain fandom communities occurs. This research explores the effects of an event-induced sudden influx of newcomers into online fandom communities. Specifically, it studies the user behavior of the newcomers and existing fans and also t e pattern of interaction among different types of fans. Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: WU Yifeng / QFIN Course: UROP1100, Summer An image conveys semantic information to the recipient, but there is a perceptive gap between what image the presenter wants to convey and how image the recipient tends to interpret. In the context of online medical crowdfunding, this gap befuddles campaign owners when they are choosing a cover image to leave first impression that has persuasive power on potential donors. This research attempts to model recipient’s first impression with an image through deep learning modelling. This paper reports research assistant’s contributions and takeaways throughout different stages of this research, including exploratory data analysis works that visualize collected data, survey on references that justify certain assumptions and operations made during the research, and data cleaning for better training results. 133 School of Engineering Department of Computer Science and Engineering Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: XU Boran / COMP Course: UROP1100, Summer Computational notebooks are known to be the ideal environment where data science studies are carried out, but a slide deck is still necessary if one wishes to present his/her study to others. Although many efforts have been sought to guide and streamline the presentation of data science works, automated slide generation remains an unexplored field. We propose nb2slide, a system that convert a Jupyter notebook into a presentation slide deck. With the help of machine-learning models and a searching module, the system essentially fills in an existing template using the contents from the notebook. The slide deck generated by the system is satisfactory. Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: ZHOU Siyuan / DSCT Course: UROP1100, Summer Online medical crowdfunding platforms (OMCPs) such as GoFundMe1 , Qingsongchou and Shuidichou, serve as online donation platforms where people can seek financial support. Therefore, attracting users' attention to make donation is crucial for campaign sponsor, in which first impression plays a very important role. In our project, we want to answer below questions: RQ1) Can users form consistent first impressions, measured by consequent motivation and projected behavioral intention given images in OMCPs? RQ2) How are impression and donation intention correlated? How people comment differently for low & high donation intention? RQ3) What are the visual features that contribute the most to the first impressions of images in OMCPs? And RQ4) Can we evaluate medical images with the dimension of impressions? My work includes: 1) Crawling down information including images, date, title, tag, location, text from website GoFundMe in order to form a dataset. 2)Doing data cleaning after collecting information in order to eliminate difference caused by spatial and cultural differences when people are rating these pictures afterwards. 3) Using Microsoft Computer Vision API2 and Amazon Rekognition3 to extract features from images in order to form a dataset and help answer RQ3. 4) Filtering out noisy ratings automatically using python based on some criteria from collected image ratings in order to improve regression result. 5) Doing regression analysis to answer RQ2. Gaining Insights from Noisy Data: Designing Transparent System for Online Decision Making Supervisor: MA Xiaojuan / CSE Student: CHEN Zixin / DSCT Course: UROP2100, Summer AutoML is now gaining increasing attention in academia and industry. It automates the selection, composition and parameterization of machine learning models. In this project, we focus on the data scientists’ opinions towards the feature engineering process in AutoML and conduct a user study to analyse people’s attitudes toward both automated and human-based feature engineering. Based on the analysis result, our further work is to provide an approach to facilitate automated feature engineering in data scientists’ works and promote the human-AI synergy. 134 School of Engineering Department of Computer Science and Engineering Handling User Challenge in Human-Agent Interaction Supervisor: MA Xiaojuan / CSE Student: LIU Dingdong / COSC Course: UROP1100, Spring With the flourishing of online shops, service chatbots are actively involved in processing purchasing requests by customers, which inevitably leads to a challenge: how should chatbot behave when misunderstanding customers’ instructions or even challenged by customers. Instead of simply apologizing or sidetracking customers, this study explores the possibility of using self-mockery to handle user challenges. Besides, we proposed a novel method to generate self-mockery sentences based on the characteristic mapping. Up till the completion of this report, the project is under the pilot study stage. A more throughout analysis of the role of xcelf-mockery will be performed after large-scale experiments. Handling User Challenge in Human-Agent Interaction Supervisor: MA Xiaojuan / CSE Student: LIU Yuzhi / MATH-AM Course: UROP1100, Spring It is necessary for researchers to have the ability to read papers effectively throughout their academic lives. However, this can be difficult for researchers who are new to their fields. For instance, new researchers may not be able to question the papers and find the appropriate answers, which is a critical process in paper reading. Human-computer interaction (HCI) researchers have explored several ways to facilitate the reading experience. Based on the previous methods, we propose an interactive agent that can generate questions automatically based on the users’ browsing behavior. Several question generation techniques are discussed in this report, and our future work on question generation will focus on the template-based method. Handling User Challenge in Human-Agent Interaction Supervisor: MA Xiaojuan / CSE Student: YU Yiduo / COMP Course: UROP1100, Spring This is a report of professor MA Xiaojuan’s UROP project Handling user challenge in human-agent interaction in Spring 2021 directed by Ph.D. Liu Chengzhong. In this project, I am instructed and guided to dig into both the generation and design of Self-mockery. Intensive literature reviews are also done in the HCI area. This report briefly introduces the general outlines of my work during this semester. 135 School of Engineering Department of Computer Science and Engineering Handling User Challenge in Human-Agent Interaction Supervisor: MA Xiaojuan / CSE Student: LIU Yuzhi / MATH-AM Course: UROP2100, Summer It is necessary for researchers to have the ability to read papers effectively throughout their academic lives. However, this can be difficult for researchers who are new to their fields. For instance, new researchers may not be able to question the papers and find the appropriate answers, which is a critical process in paper reading. Human-computer interaction (HCI) researchers have explored several ways to facilitate the reading experience. Based on the previous methods, we propose an interactive agent that can generate questions automatically based on the users’ browsing behavior. Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: GUO Bingcan / DSCT Course: UROP1100, Fall Community Question Answering (CQA) platforms have been an important source of knowledge sharing. Users can raise(answer) questions and rate other’s questions(answers) on the websites. By doing so, they will receive reputation points and badges to signify their contribution. However, while quality questions attract constant discussion, some other questions, even with similar content, are ignored. Hence, we want to find what makes an effective question that receives answers soon, scores high or somehow contributes to the CQA community. By carrying out quantitative data analysis, we hope to uncover traits of effective questions and propose possible guidelines for asking questions more efficiently. In this report, data extracted from the Stack Overflow website are studied. We summarize statistical features in table format and carry out hypothesis testing. Preliminary findings and plans for the next phase will be discussed eventually. Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: ZHOU Yuran / COGBM Course: UROP2100, Fall The stack overflow platform has become the CQA platform with the highest traffics and attention. After researching and extracting data from the top three computational and non-computational CQAs from the previous project, a decision is made to focus more on Stack Overflow from more detailed features and further implications. With new tools and methods implemented, we are now able to mine and extrapolate more information from this website to perform further analysis. As a part of a larger research, this project will mainly focus on the implication of the number of verbs within the question title and how it may affect the quality of the question via multiple standards. In this report, the means of data extraction and manipulation will be explained, the specific implication of verbs will be discussed, and the final recommendation of wording will be given to facilitate CQA users to raise popular questions. 136 School of Engineering Department of Computer Science and Engineering Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: ZHOU Taichang / COMP Course: UROP1100, Spring This report will introduce a developing software constituted by four major parts: facial feature collection, gaze point detection, engagement analysis, and result exhibition. The facial feature collection is done by OpenFace [1] (an open-source software generating expressive and gestural attributes), which giving out a .csv file. For the gaze point detection, we use data gained from OpenFace to develop a calculate based algorithm. Then we build a Machine Learning algorithm to train our prediction model of students' engagement during e-learning classes. The result exhibition is still under study. Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: GUO Bingcan / DSCT Course: UROP2100, Spring Community-based Question Answering forums (CQAs) are important repositories of public knowledge and powerful complement for search engines. How to benefit the sustainable development of the CQA online community has been a popular research topic. Among all the QA forums, academic CQAs are one main category, which are usually dominated by questions about objective fact and requesting professional answers. However, due to the questions’ requirement for answerer’s expertise, many users are hindered from contributing answers and involving in the discussion. Boosting user’s participation in the objective context of academic CQA questions is hence essential. Inspired by the effect of personal pronouns in advertising and relevant research in the subjectivity of CQA questions, we hope to know whether the presence of personal pronouns in academic CQA question titles will benefit the question to encourage users’ active participation in discussion and idea sharing. Our results show that the presence of the personal pronoun ’I’ in a question’s title lead to more answers, while other personal pronouns didn’t show significant improvement in question quality. The personal pronouns are not effective discussion triggers for academic CQA questions. We hope the conclusion can provide preliminary ideas for CQA question quality evaluation and user’s engagement in gamified academic CQA related research. 137 School of Engineering Department of Computer Science and Engineering Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: ZHOU Taichang / COMP Course: UROP2100, Summer As online learning is generally becoming a significant learning method worldwide, teachers are facing new challenges. One of them is that it is not easy for teachers to find students’ learning status due to the unwillingness of students’ turning on the camera and the limit of the screen size. Therefore, we are building a new assistant system distributing students’ learning status in real-time. This report will introduce how we have designed and realized the user interface for the tool mentioned in my last UROP report based on data from the backend and the design of our experiment. For the interface implementation part, we developed a web-based interface that displays the overall condition of a set of students' learning statuses, including 1) engagement level, 2) confusion level, 3) gaze concentration, 4) emotion, and 5) some informative facial behaviors. For the experiment part, we rented and deployed a server running data collection and data integration. On the one hand, the server obtains data from each student. On the other hand, the server computes the overall statistical data and transfer it to the interface. Paper-based VR/AR Interaction Supervisor: MA Xiaojuan / CSE Student: CHAO Chung-chi / COGBM Course: UROP1100, Fall Augmented reality (AR), a tool to bring virtual objects to a real environment, has been applied in various settings such as e-shopping experience enhancement, entertainment, and assistance for people with disabilities. Applicable devices includes AR Head-Mounted Displays (HMD), AR glasses, and more. Common ways of interacting with such systems are through controllers, gestures, gaze or keyboard interactions. This study will discuss the possibility of applying voice recognition technology when commanding the systems to ease the difficulty users may face when interacting with an unfamiliar system. Various speech recognition methods and aspects of considerations that could be applied in the system will also be discussed. 138 School of Engineering Department of Computer Science and Engineering Paper-based VR/AR Interaction Supervisor: MA Xiaojuan / CSE Student: TSANG Cheuk Nam / COMP Course: UROP1100, Fall Augmented reality is one of the new frontiers for bridging the digital world and the real world. One of the more immersive approaches utilizes an optical see-through head mounted display as an extension of paper-based interactions, but this approach offers limited immersion due to blending of light from both the projected image and environmental source which can lower immersion and legibility. In this study, we summarized the common limitations of optical see-through displays and designed a color compensation algorithm to rectify the effect of color blending on color accuracy. We further extend the comparison between potential approaches of color correction and their computational cost when implemented with constrained resources on mobile AR devices such as the Microsoft HoloLens. Predicting Student Performance on an E-Learning Platform Supervisor: MA Xiaojuan / CSE Student: ZHANG Yuanhao / COMP Course: UROP1100, Fall Community Question Answering websites (CQAs) such as Stack Exchange, Quora, Zhihu, and Answers.com serve as online knowledge hubs where people can ask their questions and answer one another. Therefore, attracting users' attention to make more and better contributions is crucial to ensure these websites' sustainability. Gamification, the use of game design elements in non-game contexts, is one way to attract users. In this paper, we investigate Winter Bash (WB), a seasonal gamification scheme on Stack Exchange, to understand the barriers that impede users from fully engaging with this type of gamification. We run a thematic analysis on text-based contents from Stack Exchange and summarize the main barriers. Finally, after constructing a coding scheme, we try to decipher it and provide some suggestions for CQA users and website to avoid or mitigate such barriers. Predicting Student Performance on an E-Learning Platform Supervisor: MA Xiaojuan / CSE Student: ZHANG Yuanhao / COMP Course: UROP2100, Spring Stack Overflow(SO) is an online Community Question Answering website (CQA) that provides a platform for programmers to exchange knowledge in the computer science field. In our study, we conduct a quantitative research to explore the relationship between the attractiveness of a post on SO and three different features of its title: title length, the number of punctuation marks in the title, the frequency of less common words in the title. With a rich dataset of 40675 posts, we define the features and run hypothesis tests, the result of which presents evidence that all of the three title features are correlated with the posts' attractiveness. Further, we find that linear regression is the best model to predict the attractiveness of a post based on the three features. 139 School of Engineering Department of Computer Science and Engineering Predicting Student Performance on an E-Learning Platform Supervisor: MA Xiaojuan / CSE Student: ZHANG Yuanhao / COMP Course: UROP3100, Summer Nowadays, gamification, the use of game design elements in non-game contexts, is a widely used measure for some learning applications (e.g., Duolingo) to stimulate users’ interest and passion for learning and to increase their learning efficiency. Nevertheless, gamification is not always successful as it sometimes distracts users from their initial goal, learning. To get the bottom of this phenomenon, we conduct qualitative research by using Duolingo as a case study. With an extensive dataset generated from the Duolingo forum over the last nine years, we run a content analysis and summarize the main ramifications of such misusing behavior and the reasons behind it. Moreover, we try to decipher them and make some reasonable suggestions to resolve or undercut these issues. A Study on the Principles and Implementations of Array Databases Supervisor: NG Wilfred Siu Hung / CSE Student: HAU Yiu Tong / COMP Course: UROP1100, Fall Data Versioning in database systems is a crucial feature for conducting scientific research. As a popular choice for large-scale data management in scientific communities, distributed array database systems require an effective storage mechanism for data versioning. In this report, we present a method for finding an optimal storage strategy for distributed array databases. First, the problem is formulated into a graph problem, then using a multilevel k-way graph partitioning scheme, version arrays are separated into different partitions representing computing clusters where similar arrays are stored together. Then, each partition is optimized to find an effective storage solution for storage space and array recreation time. Develop a Machine Learning Application on Video Data Base Retrieval Using Tensorflow Facilities Supervisor: NG Wilfred Siu Hung / CSE Student: TSE Wai Chung / SENG Course: UROP1100, Summer Semantic video retrieval is one of the most common requirements for computer users. That is, given a user query, we want to extract all the videos that are semantically related to the query, without requiring the user to do any preliminary tagging on the videos. This report presents an approach to tackle this task. With the support of TensorFlow, we applied transfer learning to state-of-the-art convolutional neural networks Xception and I3D Kinetics, allowing them to interpret videos and automatically assign them with tags. Then, word embedding semantic search techniques are applied to multi-match user queries to auto-tagged videos, achieving the task of semantic video retrieval. 140 School of Engineering Department of Computer Science and Engineering Acceleration Techniques for Real-Time Rendering Supervisor: SANDER Pedro / CSE Student: DEMIRER Caner / COMP Course: UROP2100, Fall This work aims to improve the object-wise reprojection cache refresh method explored in the previous term. To this end, Variable Rate Shading is introduced to the rendering flow in order to have more granularity in the choice of cache refresh quality while making use of a recent GPU feature for additional performance gains. Currently the implementation of the latest ideas and their testing is still ongoing. There are some early findings that encourage further exploration of the technique. This paper elaborates on the method that is developed, the current status of the project, and ideas to further explore in order to further improve the results. Acceleration Techniques for Real-Time Rendering Supervisor: SANDER Pedro / CSE Student: HU Benran / DSCT Course: UROP1100, Spring Interactive demonstrations are effective teaching-assisting tools for introductory computer graphics courses. However, current existing demonstrations usually focus on few individual topics, while demonstrations for some concepts are rare. To address the issues, this project develops an interactive platform to illustrate the major concepts and techniques of rendering covered in introductory graphics courses. A unified scene is used to enable comparisons and experiments on combinations of different techniques. The project is developed based on an existing demonstration tool Metarenderer. Extended features include multiple fully configurable and controllable cameras; visualizations of culling, z-buffer and SSAA; demonstrations with adjustable parameters on texture mapping, procedural textures, light sources, shadow mapping, microfacet BRDF and precomputed radiance transfer. Acceleration Techniques for Real-Time Rendering Supervisor: SANDER Pedro / CSE Student: HU Benran / DSCT Course: UROP2100, Summer Virtual reality applications, due to their inherent requirements on latency and mobility, usually pose different challenges comparing to traditional applications in terms of rendering. This survey provides some insights into these issues and existing approaches. First, we introduce different forms of VR applications and the respective challenges exposed to them. Then, we give a brief overview of several types of solutions tackling the issues and related work. Despite the popularity and fast pace of development of this field, many unsolved technical problems remain. We finish by summarizing some limitations of these techniques and possible future directions. 141 School of Engineering Department of Computer Science and Engineering Efficient Algorithms to Process Gigapixel Images Supervisor: SANDER Pedro / CSE Student: XU Peiyu / COMP Course: UROP3100, Fall We present a method for rendering acceleration by exploiting the temporal coherency between frames. This project is an optimization of existing algorithm that caches certain information from previous frames, and project them when rendering the next frame. In its simplest form, a regular refresh of the cache is required, which is often expensive. Such refresh cost shall be amortized in application. We managed to propose an algorithm for intelligently refreshing the cache, taking information for each object into consideration. Commonsense Reasoning with Knowledge Graphs Supervisor: SONG Yangqiu / CSE Student: XIANG Tianqi / COSC Course: UROP2100, Fall Presently reported is a progress report of a UROP research trying to accomplish the task 4 of SemEval 2021, a competition in natural language processing area. The project is going to firstly prepare some knowledge of this area. Then it is going to test and evaluate the provided baseline models. Finally, it is going to improve and implement some other models to deal with the tasks and evaluate the effectiveness of the new models. This progress report will show the principle of the operations, the procedure of this research, the progress of this project and the future plan. Commonsense Reasoning with Knowledge Graphs Supervisor: SONG Yangqiu / CSE Student: CHOI Sehyun / COMP Course: UROP1100, Spring Commonsense Knowledge Bases (CSKB) provides a venue to let the machine to understand commonsense knowledge, which are often omitted in human generated text. Such knowledge bases are costly to build in that requires extensive human annotations, and therefore attracted a research direction of automating the process. However, reasoning over such knowledge bases is a difficult task. Previously, CSKB completion aims to introduce missing edges between known entities, whereas CSKB population proposes to introduce both new entities and links between them from other sources. An effective high-quality dataset for the task is proposed, along with novel graph reasoning models that outperformed baselines from previous works. 142 School of Engineering Department of Computer Science and Engineering Commonsense Reasoning with Knowledge Graphs Supervisor: SONG Yangqiu / CSE Student: ZHAO Jiachen / COMP Course: UROP1100, Spring Gamification is a method to motivate users by applying game elements to non-game situations. Building a satisfying gamified platform has advantages compared to ordinary ones, for example, more user activities and better outcomes. However, an unsuccessful gamification may have countereffects, and waste the efforts paid on it. In this study, we conduct a qualitative research to do sentiment analysis on the posts from Meta Stack Overflow. We will use a pre-trained natural language processing (NLP) model Bidirectional Encoder Representations from Transformers (BERT) to analyze the post texts to see whether gamification of Stack Overflow is satisfying and find out a pattern between questions and answers. Commonsense Reasoning with Knowledge Graphs Supervisor: SONG Yangqiu / CSE Student: WANG Weiqi / COSC Course: UROP1100, Summer Commonsense knowledge is crucial for artificial intelligence machines to understand natural language. While learning commonsense knowledge with the help of knowledge graph can improve the performance of deep learning models, most of the existing commonsense datasets use (ℎ, 𝑟, 𝑡) formed CSV file as the dataset carrier, and the corresponding knowledge cannot be easily matched in the knowledge graph. In this project, we aim to align GLUCOSE, a commonsense dataset annotated by humans based on the ROCstory, to ASER, a large-scale eventuality knowledge graph, to investigate how much commonsense knowledge can be grounded in the eventuality knowledge graph. This resource can also serve as the preparation work for future studies on commonsense knowledge graph reasoning. Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: ORAZALIN Alibek / MATH-CS Course: UROP1100, Fall Given two vertices s and t in a weighted graph, the shortest distance/path needs to be found efficiently where weights of edges may change over time. This query has many practical applications such as finding shortest path between two points on a map where time travelled for each part of the road may vary due to various reasons (traffic jams, weather conditions, etc.). Recent paper found an efficient way to find the shortest path in the graph in 𝑂(𝑙og4𝑛 ⋅ 𝑙og 𝑙og 𝑛 + 𝑙) time, while preprocessing takes 𝑂(𝑛 ⋅ 𝑙og2𝑛) and update takes only 𝑂(𝑙og3𝑛) time which is a significant improvement. This report presents the current progress regarding the research mentioned above. 143 School of Engineering Department of Computer Science and Engineering Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: PATUPAT Albert John Lalim / DSCT Course: UROP1100, Fall Personalized PageRank (PPR) - also known as Random Walk with Restart - is one of the most deeply studied node proximity measures in the graph data mining community. Specifically, for frequently-evolving graphs, it is typically required to recompute PPR values which immediately reflect changes, and hence, lightweight index-free approximate algorithms are preferred for this use case. The state-of-the-art approach FORA achieves both efficiency and theoretical guarantees, but its theoretical analysis and implementation details show room for improvement. Thus, this study implemented the proposed algorithm in the recent theoretical analysis improving FORA. Results of rigorous testing show that the proposed algorithm is consistently faster than FORA and other state-of-the-art algorithms. For future work, this study could motivate faster FORA-like algorithms for PPR-related queries. Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: LIU Zichen / DSCT Course: UROP1100, Spring There are various types of queries over database, and they play an important role in many areas nowadays. However, these queries usually turn out to be NP-hard, and the exact algorithms are sophisticated. This project mainly focusing on the following 3 cases: queries of the fastest path over a dynamic network; queries associated with spatial keywords; queries about spatial crowdsourcing and ridesharing. This semester, the project mainly focused on understanding relevant areas and the classic algorithms. Learning how to efficiently solve a query problem with good approximate algorithms is the first step of the research. The report will show the progress. Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: ORAZALIN Alibek / MATH-CS Course: UROP2100, Spring Given a directed weighted graph G, our aim is to answer the shortest distance, path queries by also permitting updates on the weights of edges. This problem, of course, has many applications such as finding shortest distance between two points in a map where weight of the edge or the time to pass a portion of the road changes (due to various reasons, such as weather conditions, traffic jams and accidents); finding the most optimal transaction between currencies (by considering logarithms of ratios). This abstract summarizes the progress on optimizing the study of fast path and distance queries on a dynamic spatial network. 144 School of Engineering Department of Computer Science and Engineering Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: PATUPAT Albert John Lalim / DSCT Course: UROP2100, Spring Recommender systems have been serving as increasingly important navigation tools tackling information overload present in the digital economy. As an emerging paradigm of this field, anonymous session-based recommender systems model short-term yet dynamic user preferences solely based on anonymous sessions and interactions. This report reviews the different approaches for anonymous session-based recommendation, highlighting representative algorithms. Afterwards, several weaknesses with current methods are discussed most notably, readily available timestamps of user actions are often discarded despite carrying meaningful information. Ultimately, this paper proposes a timestamp-aware graph neural network addressing these problems. Future direction of such work includes empirical analysis of the proposed model, theoretical analysis of the training and prediction algorithm, and extension of the approach for online data stream environments. Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: LAU Yan Hei / SENG Course: UROP1000, Summer The following report consists of a summary of the knowledge acquired during the UROP1000 course, under the guidance of the instructor. The courses studied include COMP2011 Introduction to Object-oriented Programming, COMP2012 Object-Oriented Programming and Data Structures and COMP3711 Design and Analysis of Algorithms. The contents below serve to demonstrate that the student has achieved a thorough understanding of the materials presented in the above courses. Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: LIU Zichen / DSCT Course: UROP1000, Summer Collective Spatial Keyword Queries (CoSKQ) aims to find a set of spatial objects that collectively satisfy keywords requirements. However, the queries studied so far generally focusing on “covering” the requirement. Among these query keywords, some of them may be of particular interest. We define the Interest-Aware and Distance-Constrained Collective Spatial Keyword Query such that the returning set not only covers the keywords but also carries as many objects of interest as possible under a distance threshold. Specifically, we study 3 variants of this problem based on different distance functions for the set. In this paper, we prove that it is NP-hard to solve the query with any constant factor and provide both exact and bi-approximation algorithms. 145 School of Engineering Department of Computer Science and Engineering Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: SU Hong / DSCT Course: UROP1000, Summer Co-location spatial data mining is a popular topic currently, which can mine some implicit information behind it. Since spatial data mining is very important, temporal-spatial data mining will mine more information. This report will first briefly introduce my immature proposed problems concerning spatial co-location data mining with dynamic time, and then present the related work and preparations I made. After that, some attempts to tackle the problems including some challenges to solve in the future will be shown. Lastly, some reflections will be made to indicate what I have learnt in this summer urop. Efficient Queries over Database Supervisor: WONG Raymond Chi Wing / CSE Student: WEI Yuanjing / DSCT Course: UROP1100, Summer Session-based recommendation system, SBRS for short, has become a trending topic in recent years and has been widely applied in various domains such as e-commerce and online media platforms. However, although many related studies have been conducted and high practical value has been approved in real-life applications, this area has been relatively under-explored. In this report, I will show some widely applied methods for SBRS, summarize the relevant studies for each of the methods, and introduce an existing problem that represents a direction in which further research can be conducted. Knowledge Discovery over Database Supervisor: WONG Raymond Chi Wing / CSE Student: LIANG Houdong / DSCT Course: UROP1100, Fall T-Music is an algorithm that composes a series of musical notes from input lyrics using frequent pattern mining and other data mining techniques. T-Music Web System is a maintainable and scalable system that deploy the usage of the algorithm. It establishes an online interface and allows users to interact with T-Music. However, the implementation of the web system has not been completed yet and there is some future work to do. This project aims to equip the group members basic front-end web design skills as well as to implement some new features based on the original web system. 146 School of Engineering Department of Computer Science and Engineering Knowledge Discovery over Database Supervisor: WONG Raymond Chi Wing / CSE Student: WU Lijia / COMP Course: UROP1100, Fall Interactive regret minimization is often used to retrieve the favourite subset of the user from a large database containing many tuples. To achieve strongly truthful results, users are required to choose his/her favourite tuple out of a small subset from the database for a small number of times. Specifically, an unknown utility function is introduced in the k-regret minimization to represent the preference of the user. With such utility function, the k-regret minimization query can reduce the regret ratio of the user. This report mainly focuses on the impacts of different forms of utility function in strongly truthful k-regret minimization query with user interactions, which includes Linear, Quadratic and Gaussian form of utility functions. Knowledge Discovery over Database Supervisor: WONG Raymond Chi Wing / CSE Student: DO Thuy Trang / DSCT Course: UROP2100, Fall The recommender system has played an important part in many online services such as e-commerce. To provide customers the most related item or service that they are likely to click, we need to predict users’ next action based on the sequence of their previous actions in a given session. From the paper: 'Session-based Recommendation with Local Invariance', we could find two significant ideas: order of actions in a given session has a great impact on the next click of customers, and the definition of 'short sub-sequence' when the products have high similarity. Based on the ideas of this paper, in this report, we propose two methods to improve the accuracy of prediction, which is Recurrent Neural Network and Bidirectional Recurrent Neural Network. Research on Mining Course Structure Supervisor: WONG Raymond Chi Wing / CSE Student: LIU Ziming / COSC Course: UROP1100, Fall In this UROP 1100 project, students should perform data mining tasks on all important components of courses on a given list to extract their structure and correlation among different courses. As a well-developed data mining application, T-Music Web System is a maintainable and scalable system to deploy T-Music algorithm, which is an innovative algorithm that can compose a series of musical notes from a given series of input lyrics. The group members of this project should fulfill the unfinished goals listed in the future work of the T-Music Web System Report, which requires the members to master basic front-end web programming skill in order to implement the new features in the web system. 147 School of Engineering Department of Computer Science and Engineering Research on Mining Course Structure Supervisor: WONG Raymond Chi Wing / CSE Student: HSU Shang-ling / COSC Course: UROP3100, Fall Student performance prediction is essential especially during the pandemic since automatic, personalized education has become more common thanks to the switch to online teaching mode. In this research, we aim to design a model that can accurately predict students' future performance given their performance in the past. Based on the previous progress report regarding the literature review, we have proposed a preliminary model. Since the model is based on many assumptions regarding the attributes of the data, in this report, we will examine those assumptions by answering a few research questions with experiments on our targeted dataset, which can lead to possible improvement of the model design. Research on Mining Course Structure Supervisor: WONG Raymond Chi Wing / CSE Student: ZHOU Jiaxi / DSCT Course: UROP1100, Spring For more than a decade, various recommendation systems have been developed and widely applied in multiple internet areas. As the problem of information overload heating up, internet users' demand for information is increasing, and recommendation systems are playing an increasingly important role in the digitalization of various fields. Among all kinds of recommendation systems, the session-based recommendation has been becoming increasingly favored by both academia and industry applications (Gorakala & Usuelli, 2015). This article will review the major recommendation systems and especially focus on some recent discoveries in the session-based recommendation. Research on Mining Course Structure Supervisor: WONG Raymond Chi Wing / CSE Student: LIU Ziming / COSC Course: UROP2100, Spring In this UROP 2100 project, students should perform data mining tasks on all important components of courses on a given list to extract their structure and correlation among different courses. As a well-developed data mining application, T-Music Web System is a maintainable and scalable system to deploy T-Music algorithm, which is an innovative algorithm that can compose a series of musical notes from a given series of input lyrics. Last semester we have learned JavaScript in order to read the code of the project and propose some ideas for the future work of the project. Before realizing it, I followed the advice from the project developer to start from learning JavaScript React. Besides, since I have finished learning of data structure and algorithm, I read papers that I have thought to be incomprehensible before. I will demonstrate what have learn in two parts. 148 School of Engineering Department of Computer Science and Engineering Research on Mining Course Structure Supervisor: WONG Raymond Chi Wing / CSE Student: HSU Shang-ling / COSC Course: UROP4100, Spring Student performance prediction is a crucial task in education. However, many previous studies suffer from 1) the insufficient utilization nor fusion of the structural and temporal information, as well as 2) the lack of explainability. To this end, we address the problem by proposing Dynamic Heterogenous Attention Network (D-HAN), formulating student performance prediction as a link prediction problem in dynamic heterogeneous graphs, like the link prediction in the Author-Paper-Venue graphs. D-HAN is mainly composed of heterogenous graph neural network modules for structural message passing, multi-head self-attention modules for temporal message passing, and the structural-temporal aggregation module for link prediction, which yields the performance prediction along with the importance scores of the neighbor nodes and thus provides some explainability. Real-time 3D Scene Reconstruction from End-to-End Deep Learning Supervisor: XU Dan / CSE Student: KIM Jaehyeok / COMP Course: UROP1000, Summer 3D scene reconstruction is a task to render 3d details of a certain scene such as depths from a set of 2D images with different viewpoints. Real-time scene reconstruction is an important task in various real-life applications such as self-driving, AR/VR, robotic navigation, and intelligent manufacture. Recently, a model representing a scene as Neural Radiance Fields (NeRF) has achieved state-of-the-art results on the 3D scene reconstruction task. Furthermore, NeRF-based variants have actively been researched to improve the drawbacks of the base model and achieve new state-of-the-art in different metrics. Although NeRF and its variants still have unresolved drawbacks, it is a promising and interesting research topic in the computer vision community. In this report, I will review the current state-of-the-art models in NeRF-based 3D scene reconstruction and discuss my progress on different research directions. Self-Supervised Scene Depth Estimation from the Wild Supervisor: XU Dan / CSE Student: YU Mukai / ISDN Course: UROP1100, Summer In recent years, deep learning has been progressing fast in computer vision, which has a lot of practical applications, including simultaneous localization and mapping (SLAM), pose estimation, etc. However, the annotated datasets for supervised deep learning are still far from adequate, at least compared to the NLP area, and this urges the computer vision academia to further structure the network and apply some hacks into it. This report investigates a novel method that simultaneously learns depth, ego-motion, and camera intrinsic from unannotated videos recorded using monocular cameras. This approach is similar to a previous method of unsupervised learning researched and published in CVPR-2017, which only learns depth and ego-motion. 149 School of Engineering Department of Computer Science and Engineering Comparative Studies of Programming Languages Supervisor: ZHANG Charles Chuan / CSE Student: CHOI Siu Kai / COMP Course: UROP1000, Summer This paper will find out the strength and weaknesses of two programming languages C++ and Rust, by analyzing the manipulation of the container. We would evaluate deque first. The method compares the runtime of bubble sort and insertion sort with random integers loaded in a deque for both languages. The result shows that Rust has a faster runtime than C++ for both sorting algorithms in all datasets. We also observed that C++ could have more freedom on container type and iteration method. The result from the experiment could be that the iterator-based for loop in Rust allows faster iteration in sorting algorithms. Thus, it is showing that Rust has a faster performance over C++ on container manipulation. Comparative Studies of Programming Languages Supervisor: ZHANG Charles Chuan / CSE Student: HU Yicheng / COMP Course: UROP1000, Summer Rust is an emerging compiled programming language that resembles the syntax of c++, it constructs a set of “ownership” rules to resolve several issues occurring in c++ during the access, allocation, and deallocation of dynamic memory, as well as concurrency control. The rust standard library has implemented several sequential and associative containers which can be used as replacements to c++ STL containers. Programmers switching from c++ to rust might struggle to choose the most appropriate language which best suits their needs. This report aims to evaluate the performance of containers in two languages so that programmers can have an initial impression of which language they should use if they have decided the containers to be used. 150 School of Engineering Department of Electronic and Computer Engineering Department of Electronic and Computer Engineering Projects in Audio Signal Processing - A Real-Time Wavelet-Based Algorithm for Improving Speech Intelligibility Supervisor: CHAU Kevin / ECE Students: CHEN Siyu / CPEG Courses: UROP1100, Spring DINH Anh Dung / COSC UROP1100, Spring KANG Tianqu / EEGBM UROP1100, Fall UROP2100, Spring UROP1000, Summer LAW Jun Jie Johnathan / MATH-CS UROP1100, Spring SIM Yi Hang / EEGBM UROP4100, Fall WANG Binghong / ELEC UROP1100, Spring YANG Jingwen / COSC UROP1100, Spring A wavelet-based algorithm to improve speech intelligibility is reported. The speech signal is split into frequency sub-bands via a multi-level discrete wavelet transform. Various gains are applied to the sub-band signals before they are recombined to form a modified version of the speech. Dynamic range compression then follows to control the peak amplitude. The sub-band gains are adjusted while keeping the overall signal energy unchanged, and the speech intelligibility under simulated hearing loss conditions and various background interference is enhanced and evaluated objectively and quantitatively using Google Speech-to-Text transcription. For English and Chinese noise-free speech, overall intelligibility is improved, and the transcription accuracy can increase by over 80 percentage points by reallocating the spectral energy toward the mid-frequency sub-bands, effectively increasing the consonant-vowel intensity ratio. This is reasonable since the consonants are relatively weak and of short duration, and are therefore the most likely to become indistinguishable in the presence of background noise or high frequency hearing impairment. For speech already corrupted by noise, improving intelligibility is challenging but still realizable. The proposed algorithm is implementable in real-time and comparatively simpler than previous algorithms. Potential applications include speech transmission, hearing aids, machine listening, and a better understanding of speech intelligibility. Yijia Chen, Tianqu Kang, Jiakun Zheng, Yuxuan Wan, Keegan Yi Hang Sim, Eugene Chau, and Kevin Chau, “A Real-Time Wavelet-Based Algorithm for Improving Speech Intelligibility,” presented at the Acoustical Society of America 179th Meeting, online, Dec. 2020. Proceedings of Meetings on Acoustics, vol. 42, 060009 (2020). https://asa.scitation.org/doi/10.1121/2.0001407 151 School of Engineering Department of Electronic and Computer Engineering Projects in Audio Signal Processing - A Multi-Band Compression Approach to Improving Speech Intelligibility Using Wavelet-Based Algorithm Supervisor: CHAU Kevin / ECE Student: DINH Anh Dung / COSC Course: UROP1100, Spring A paper by Yijia Chen et al. from our UROP team has explored a method for compensating the effects of hearing loss by applying a real-time wavelet-based algorithm to separate the speech signal into multiple sub-bands, which can then be amplified independently. Due to the nature of hearing loss in reducing the dynamic range of hearing differently at various frequencies, the wavelet method allows a more accurate restoration of the original signal compared to amplifying the whole speech uniformly, which may cause audio sections already loud to becoming too loud for listeners. The original paper tackled this problem by putting a speech audio with or without background interferences through a hearing loss simulator and then applying the wavelet-based modifications, using Google speech-to-text transcription for objective evaluation and optimization of speech intelligibility. This report attempts to improve the efficiency of the original algorithm by applying compression techniques on the wavelet sub-bands instead of the entire audio and then amplifying it to the maximum allowable speech energy. Projects in Audio Signal Processing - On Audio Signal Processing and Hearing Loss Supervisor: CHAU Kevin / ECE Student: LAW Jun Jie Johnathan / MATH-CS Course: UROP1100, Spring In studying hearing loss and improving speech intelligibility, fundamental notions exist to lay the groundwork for further development. This paper describes some of the key ideas involved in the UROP project, including the Fourier Transform and its variants, the Wavelet Transform, and some aspects of audiology such as loudness and loudness recruitment. We introduce the Discrete Fourier Transform along with the computationally efficient Fast Fourier Transform. We then provide motivation for and introduce the Short-Time Fourier Transform, as well as highlight its shortcomings. We also present Wavelets and the Wavelet Transform, a relatively new area of signal processing. Finally, we touch upon loudness and loudness recruitment to provide a well-rounded understanding of the subject matter related to audiology. Projects in Audio Signal Processing - The Effects of Audio Compression on Google Speech-to-Text Transcription Accuracy Supervisor: CHAU Kevin / ECE Student: WANG Binghong / ELEC Course: UROP1100, Spring This report studies the relationship between the degree of speech audio compression and the accuracy of Google Speech-to-Text transcription. Lossless speech examples in English, Chinese, French and German by both male and female speakers are converted into MP3 format and compressed at different bit rates, and the corresponding Google Speech-to-Text transcriptions are compared with the correct scripts. Fluctuations of transcription accuracy are observed at low bit rates, whereas the accuracy generally improves and maintains a high level when the bit rate increases. It is also observed that Google transcription can stop at times regardless of the bit rate, resulting in incomplete transcription and poor accuracy. Overall, Google Speech-to-Text converter is highly reliable for compressed speech audio files at a relatively high bit rate. 152 School of Engineering Department of Electronic and Computer Engineering Projects in Audio Signal Processing - A Summary on the Simulation of Sensorineural Hearing Loss Supervisor: CHAU Kevin / ECE Student: YANG Jingwen / COSC Course: UROP1100, Spring This report reviews the operating principles of the hearing loss simulator in our previous work and presents a summary of additional features from others’ work that we have not included. Typical consequences of sensorineural hearing loss are 1) an elevated threshold of hearing, 2) an abnormal loudness growth with sound level, 3) a reduced frequency selectivity, and 4) an unaffected just noticeable difference in intensity. While our current simulator only models threshold elevation and loudness recruitment, the latter two effects can be incorporated by applying the signal processing schemes proposed here. These upgrades should enable a more accurate modelling of hearing loss, from which we can develop a better compensating algorithm for the hearing impaired. Nanostructured Chemical Sensors and Sensor Systems Supervisor: FAN Zhiyong / ECE Student: CAI Yuxiang / ELEC Course: UROP1100, Summer Currently, robots can not only see with the help of image processing technology but also hear with voice processing technology. Chemical sensor will enable the mechines to obtain a sence of smell. This technology will be widely applied in people's daily life. For example, detecting the dangerous gases at factories and adjusting air circulation at home both need a sensitive electronic nose. Robots with chemical sensors can also replace the detection dogs in the future. We choose AAO (anodic aluminum oxide) as the material of sensors. The reasons are as follow: First, the surface to volume ratio of AAO is high. Second, AAO is a material with a stable performance. Finally, the cost of AAO is acceptable. The direction where we are exploring is to balance the precision, cost and volumn of chemical sensors. Nanostructured Chemical Sensors and Sensor Systems Supervisor: FAN Zhiyong / ECE Student: YANG Zheng / ELEC Course: UROP1100, Summer For this summer, from knowing nothing about perovskite and its applications and groupmates’ work to have a brief knowledge about this filed and how their work conduct, I have learned many things. For the first two months I mainly learn about how to make the anodic aluminum oxide. And the last month I start to learn some hardware knowledge. So, for this report, I mainly talk about the process to make the AAO and elaborate each details and what I have learned from this process. As the end of this report, I will briefly talk about my feelings after a summer’s study. 153 School of Engineering Department of Electronic and Computer Engineering Artificial Intelligence Methods for Medical Videos Supervisor: LI Xiaomeng / ECE Student: TSANG King Hoi / COMP Course: UROP1100, Summer This report mainly reviews the basic concepts of deep learning and various network architectures used in the context image and video analysis that was explored throughout the course of this project. Some of the concepts are then used to design a network architecture to perform classification and localization tasks and evaluated using the ESAD and MESAD dataset. Concepts discussed includes convolutional neural networks, residual neural networks, the YOLO algorithm residual neural networks and Feature pyramids. These concepts will be used to explain different architectures used for image and video analysis explored throughout the duration of this project, such as the RetinaNet network and examples of semi-supervised networks for segmentation tasks. End-to-End Instance Segmentation with Transformers Supervisor: LI Xiaomeng / ECE Student: YU Yiduo / COMP XIANG Tianqi / COSC Course: UROP1000, Summer UROP1100, Summer This is a report of professor LI Xiaomeng's UROP project End-to-End Instance Segmentation with Transformers in Summer 2021. In this project, we are instructed and guided to dig into both the novel technologies of instance segmentation and the application of those networks. Some fundamental classical papers which laid a solid base for the image segmentation task are reviewed. This report briefly introduces the general outlines of our work during the summer. Graph Neural Network for Drug Discovery Supervisor: LI Xiaomeng / ECE Student: YANG Zheng / ELEC Course: UROP1100, Spring After a brief introduction of this project, it goes to the main part of this report. First for the related work parts I talk about my own understandings of the CNN and Transformer architecture after a semester’s study. Then I introduce two papers about combination of CNN and Transformer on medical images process. For this part, I mainly talk about their architecture respectively and compare their difference. The mainly part ends with the conclusions which is my feelings about this field after three months’ study. The last part is the reflection about this whole semester’s study including the problems I have. 154 School of Engineering Department of Electronic and Computer Engineering Learning with Limited Anotated Data for Medical Image Diagnosis Supervisor: LI Xiaomeng / ECE Student: GANESH Vignesh / SENG Course: UROP1000, Summer Medical image analysis has come leaps and bounds since the integration of technology into the field of medical science. Machine learning and artificial intelligence have substantially improved the efficiency of medical image diagnosis by aiding medical professionals in the detection of numerous diseases. These complex models are trained rigorously to classify images based on certain parameters and can identify minute differences between images that are indistinguishable to the human eye. Deep learning models require the datasets to be labelled based on certain classifications before it is passed through for the training and testing phase. Limited annotated data has made it difficult for researchers to carry out accurate medical image analysis in a field where the margins for error are infinitesimal. This project aims to implement and improvise AI algorithms that can be used for medical image analysis with limited annotated data. Through variants of supervised learning, the main goal is to make the algorithm more intelligent by training with unlabelled data or data with few labels, thereby improving its accuracy and effectiveness. Learning with Limited Anotated Data for Medical Image Diagnosis Supervisor: LI Xiaomeng / ECE Student: GUO Jiarong / ELEC Course: UROP1100, Summer The world faces considerable challenges in terms of eye care, according to WHO, the number of visually impaired people worldwide is more than 2.2 billion. Early intervention to detect and diagnose vision could reduce the impact on at least one billion people. Based on the fundus images in RFMiD, which were captured by a retinal specialist himself at an Eye Clinic and public screening camp, we were able to build up a model based on ResNet50, which can detect the disease of patients with underlying eye disease. By using a ResNet50, it brings greater efficiency in identifying the fundus images. Representation Learning for Graph Neural Networks Supervisor: LI Xiaomeng / ECE Student: YANG Jingwen / COSC Course: UROP1000, Summer While image shadow detection has obtained significant achievements, its detection over dynamic scenes is yet to be explored until ViSha, the first large scale dataset for video shadow detection. Compared with images, videos have an additional temporal dimension. In this report, we aim to evaluate the effects of the temporal information in video shadow detection. Specifically, we first review the current state-of-the-art video shadow detection method, TVSD-Net, which considers temporal information at video level. Then, we design a modified network that aims to enhance the image-level representations of TVSD-Net by the Swin Transformer. The experiment results show that better image-level representations hardly improve the performance, while the performance of lacking temporal information clearly drops. This inspires us to explore long-term temporal information for continuous research. 155 School of Engineering Department of Electronic and Computer Engineering Representation Learning for Graph Neural Networks Supervisor: LI Xiaomeng / ECE Student: TILNEY-BASSETT Oktarian George / COMP Course: UROP1100, Summer This project was split into two parts, which allowed me to gain exposure to two different deep learning areas of research. During the first half, I studied graph representation learning through contrastive learning. I started by studying graph neural networks and their properties, then implementing the graph convolutional network model. I then moved onto studying contrastive learning for graphs, an area of a research that is particularly important to solve the data labeling problem. I was able to understand the core frameworks and techniques used, as well as implement a few of the main models. Finally, I studied various models that use novel augmentation schemes to improve the overall accuracy, as contrastive learning frameworks need robust schemes to generate different augmented views for the training. In the second half of the project, I moved to study medical imaging. More specifically, I learnt about transformers and how they can be used in medical imaging to conduct automatic grading of diabetic retinopathy and diabetic macular edema. Overall, this research project has greatly increased my understanding of various deep learning components, as well as their applications to real-life tasks. Robust and Generalized Methods for Medical Image Analysis Supervisor: LI Xiaomeng / ECE Student: CAINE Wilbert / COSC Course: UROP1100, Summer Fundus fluorescein angiography (FFA) is commonly used in ophthalmology diagnosis as complementary image information to obtain a map of the vascular structure of the retina. However, the exogenous dye injected into the bloodstream can cause adverse harm to the patients of angiographic imaging. Despite the characteristic differences, conventional fundus images and FFA images still share most of the prominent features. This report applied a generative adversarial network (GAN)-based method to learn the mapping features from fundus images to FFA images. The result was validated and compared with existing generative networks for fundus-to-angiography synthesis using the publicly available Isfahan MISP dataset. The metrics used for evaluation include peak signal-to-noise ratio (PSNR), structural similarity (SSIM), and Fréchet Inception Distance (FID). Robust and Generalized Methods for Medical Image Analysis Supervisor: LI Xiaomeng / ECE Student: YANG Lin / COMP Course: UROP1100, Summer Deep learning methods have become a trend to solve various computer vision problems and have shown promising performance in recent years. Among the networks, nnU-Net (no-new-net), a convolutional neural network based on U-Net shows great performance on datasets with different contents. This article elaborates the features of nnU-Net and reviews the related works of nnU-Net, including CNNs and U-Nets, analyze their architecture, improvements, and drawbacks. In the last part, this article states a possible direction for the future study in the area of deep learning inspired by the hypothesis of nnU-Net, which is to develop a deep learning method that is universal and is easy to use. 156 School of Engineering Department of Electronic and Computer Engineering Development of a Drone Positioning System for Tiled Walls Supervisor: MOW Wai Ho / ECE Student: CHUNG Tsz Wa / SENG Course: UROP1000, Summer Recently, drone development is poplar worldwide. For instance, drones that could deliver food remotely and drones that could save people from drowning. The current Global Positioning System (GPS) allows drones to fly to different specific location outdoor. However, tasks such as scanning the exterior tiled wall of a building which requires precise location may not be able to achieve by GPS but extracting the tiled wall patterns. A prototype of drone positioning system for tiled walls had been developed by previous Final Year Project students. It could predict the distance between the drone and the wall, but it is not stable and slow. During the Summer UROP, we are able to optimize the system, making it more stable, and reducing the running time by 96.7% to 98.5% depending on the size of the input image. Development of a Drone Positioning System for Tiled Walls Supervisor: MOW Wai Ho / ECE Student: JIA Tongyao / ELEC Course: UROP1100, Summer Drones are suggested for applications in delivery services in recent years. Although the Global Positioning System allows the drone to fly to a specified outdoor location, it is not precise enough for a drone to identify and locate a specific window at a specified floor. To face this challenge, the HKUST team has proposed a vision positioning solution by using the patterns of tiled walls in building exteriors. However, the current Matlab-based prototype is feasible but too slow and not stable enough to be put into practice. This report aims at showing what has been done in UROP1100 this summer to improve the stability of the algorithm based on the Matlab version. An Optofluidic Biological Cell Stretcher Supervisor: POON Andrew Wing On / ECE Student: ZHU Guanhua / CPEG Course: UROP1100, Summer The standard TEM00 Gaussian beam intensity profile, along with other beam profiles of TEMmn (where m, n > 0 and m, n ∈ Ν), can impose a gradient force that incline to trap dielectric particles in the focal region. Using the TEM00 beams as traps, dielectric particles in the size range from 10 μm down to ~25 nm can be trapped in solution stably. By building up a modified microscope, letting the laser beam to pass through the objective lens and focus on the cells in microfluidic flow, we can stretch the cells and then estimate their elasticity, so that we can indicate the viability of the cell. This progress report illustrates the setup of such single-beam biological cell stretcher, the theory behind the critical components of the setup, and demonstrate the actual application using the device. 157 School of Engineering Department of Electronic and Computer Engineering Automated Test Equipment for Magnetic Memory and Sensors Supervisor: SHAO Qiming / ECE Student: TONG Zihan / ELEC Course: UROP1100, Fall For the past decades, research has been showing that magnetic random-access memory (MRAM) based on emerging materials is promising for applications with high requirements for data storage and access due to its prominently high access speed, low power consumption and almost infinite endurance, etc. The storage element of such a novel memory device is magnetic tunnel junction (MTJ). To perform further circuit design and device development based on MRAM, a compact model of MTJ is supposed to be developed and optimised. During the project period, the development of emerging memory devices and designs of MTJ are reviewed. A reproduction of the compact model design for voltage-gated spin-orbit torque magnetic tunnel junction (SOT-MTJ) is conducted. Various simulations based on the model constructed are also performed and discussed. Automated Test Equipment for Magnetic Memory and Sensors Supervisor: SHAO Qiming / ECE Student: GUO Jiarong / ELEC Course: UROP1100, Spring The memristor, as the fourth electrical fundamental element, has great prospects on brain-like system and new generation of computer’s CPU. Thus, testing for some basic properties of material that build on 10 nm gold base electrode, and test Hall Effect of this chip. This project is related to design a system for testing memristor. During the experiment, by using different thickness of chiral 2D Perovskitesfor Hall measurement, the testing result of chiral perovskites has great electronic properties and compare to system measurements. Designing a system for testing the chips efficiently is still on progress. So, the report is mainly focus on the process of testing the chips. Automated Test Equipment for Magnetic Memory and Sensors Supervisor: SHAO Qiming / ECE Student: TONG Zihan / ELEC Course: UROP2100, Spring Recent theories and experiments have been demonstrating that besides the binary-like non-volatile magnetization switching, perpendicular magnetic tunnel junctions (MTJs) based on spin-orbit torque switching may also show analogue-like behaviours, making them one of the possible candidates for solid-state artificial synapses and neurons, which are the fundamental building blocks of the next-generation hardware of neuromorphic computing. This project aims at developing a compact model for the multidomain MTJ devices based on the Preisach model of hysteresis, which treats the ferromagnetic material as the combination of parallel and independent hysterons with respective switching thresholds, and thus addressing their properties of multiple intermediate states and the issue of thermal stability. 158 School of Engineering Department of Electronic and Computer Engineering Automated Test Equipment for Magnetic Memory and Sensors Supervisor: SHAO Qiming / ECE Student: TONG Zihan / ELEC Course: UROP3100, Summer Recent theories and experiments have been demonstrating that besides the binary-like non-volatile magnetization switching, perpendicular magnetic tunnel junctions (MTJs) based on spin-orbit torque switching may also show analogue-like behaviours, making them one of the possible candidates for the building block of the next-generation solidstate devices of neuromorphic computing. This project aims at developing a compact model for the multi-domain MTJ devices based on the Preisach model of hysteresis, which treats the ferromagnetic material as the combination of parallel and independent hysterons with respective switching thresholds or switching probability curves. Model verification based on various experiment data is also performed and discussed. Circuit Emulation of Bio-Inspired Dynamic and Topological Quantum Systems Supervisor: SHAO Qiming / ECE Student: ZHENG Yueyan / CPEG Course: UROP1000, Summer Because of the high demand for processing the temporal signal, a kind of RNN-based framework is created and whose name is reservoir computing. Reservoir computing became popular based on its lower training cost compared with the traditional RNN. This work is based on the paper- Dynamic memristor-based reservoir computing for high-efficiency temporal signal processing (Zhong et al., 2021), aimed to detect and test the performance of a dynamic memristor-based reservoir computing which can highly efficient and accurate processing temporal signals. The entire study will be divided into three tests to test the performance of this reservoir, including waveform classification, spoken-digit recognition, and time-series prediction. We hope to develop the temporal signal processing performance of dynamic memristor-based reservoir computing. Circuit Emulation of Bio-Inspired Dynamic and Topological Quantum Systems Supervisor: SHAO Qiming / ECE Student: LO Yuk Ho / PHYS Course: UROP1100, Summer Recently, it was found that electric circuits could be used to emulate bio-inspired dynamic and topological quantum systems due to various available circuit components with different characteristics. Besides, it is more convenient if people can use simulation software to emulate their desired systems. In this project, LTspice, a high-performance SPICE simulation software, is used to emulate some topological circuits proposed by the literature ”Topolectrical Circuits”. The targeted circuits are Su-Schrieffer-Heeger (SSH) circuit and the Graphene circuit, which is a higher dimensional circuit compared to SSH circuit. From the simulated circuits, some results are obtained and used to compare with those mentioned in the literature we refer to. 159 School of Engineering Department of Electronic and Computer Engineering Circuit Emulation of Bio-Inspired Dynamic and Topological Quantum Systems Supervisor: SHAO Qiming / ECE Student: ZHENG Wenhan / SENG Course: UROP1100, Summer This UROP program has a task to reproduce a memristor emulator using electronic components. In this paper, two memristor emulator is being tested. The first one is reproduced and tested with fatal problems. Various methods including changing the input, loosening the precision restriction and changing the configuration of the circuit are used. However, it cannot give acceptable output and failed to show the hysteresis loop. Using the experience in building and testing the first model, the second model proposed in the reference of the first model is constructed and tested. By changing the input frequency and modifying the circuit, the output is much more preferable and the plot shows the symbolic hysteresis loop of the memeristor. Compact Models for Circuit Design Supervisor: SHAO Qiming / ECE Student: FANG Yidong / ELEC Course: UROP1100, Fall Spin-transfer torque magnetic tunnel junction (STT-MTJ) is an emerging field for nonvolatile memories because of its fast speed, low power and perfect endurance. In this work, the STT-MTJ basic properties has been simulated through MATLAB such as the MTJ conductance, the tunneling magnetoresistance, switching probability and critical current switching via Brinkman physical model and Slonczewski model. From the results, we can find the that the conductance of MTJ is almost constant when it is in parallel state (P state) and has a quadratic character when in anti-parallel state (AP state), also the current has a more linear feature with the bias voltage when the magnetic tunnel junction is at the parallel state and non-linear when it is at anti-parallel state. The simulation results show great consistency with the experimental results from the two models mentioned above and the paper from Beihang University. Based on these characteristics, magnetic tunnel junction has becoming one of the most promising spintronic devices in academic world and is very predicting when it is used in some nonvolatile memories such as magnetic Spin-transfer torque random access memory (STT-MRAM). Compact Models for Circuit Design Supervisor: SHAO Qiming / ECE Student: NUGRAHA Ferris Prima / CPEG Course: UROP1100, Spring The objective of this research is to develop a compact model which accurately predicts the behavior of stochastic Magnetic Tunnel Junction (MTJ) with respect to the factors including external magnetic fields, anisotropy types, and the energy barrier of the nanomagnet. This project investigates the previous research papers and attempts to confirm the correlation between the relaxation time with the factors through numerical simulations and equations. This would be critical in future hardware implementation including in the non-volatile magnetic RAM (MRAM) and probabilistic computing applications along with the Binary Stochastic Neuron (BSN) in machine learning. Several phenomena to be studied involves the spin dynamics described in the Landau-Lifshitz-Gilbert (LLG) equation, Spin-Transfer Torque (STT), Spin-Orbit Torque (SOT), and stochastic thermal noise. 160 School of Engineering Department of Electronic and Computer Engineering Compact Models for Circuit Design Supervisor: SHAO Qiming / ECE Student: XU Zimo / CPEG Course: UROP1000, Summer In this report, I proposed a physics-based compact model for the three-terminal VGSOT-MTJ device with the AFM/FM/Oxide structure, which includes the models written in MATLAB. Then I investigate the impact of the VCMA effect on the field-free SOT switching through the figures produced by MATLAB. Equipment Automation for Remote Experimentation Supervisor: SHAO Qiming / ECE Student: XING Hexiang / ELEC Course: UROP1100, Fall This project named equipment automation for remote experiment aims at developing a software platform to control laboratory apparatuses remotely. In order to achieve this function, a software tool called Labrad, a platform that can divide a complex project into simple, manageable tasks that are operated by different server modules, is used. Labrad allows users to communicate with different apparatuses directly and sufficiently through a computer, based on python programming. The report shows the basic structure and data flow in Labrad system, how to obtain data from a magnetic sensor, and the method to control a stepper motor by computer programming. Equipment Automation for Remote Experimentation Supervisor: SHAO Qiming / ECE Student: LIU Hanmo / CPEG Course: UROP1100, Spring It is known the computer algorism can do the logic control based on the input of the system. There is a balance between algorism and building the circuit. In our study, we build a Hall circuit and managed to do the electronic measurement automatically with electrical characteristics. Equipment Automation for Remote Experimentation Supervisor: SHAO Qiming / ECE Student: WANG Peiqi / ELEC Course: UROP1100, Spring A wide variety of devices and systems based on emerging materials have an exciting future in fields like AI and IoT. However, it is very time-consuming and sometimes even impossible to control or test them in person with testing equipment. It is therefore meaningful to develop a platform that can achieve the objective of “Equipment Automation for Remote Experimentation”. Apart from the controlling software, a hardware system is needed to control all apparatuses remotely for intelligent investigation. We survey the existing ADC and DAC circuits, and simulate several circuit models using LTspice. These circuits can later be used as controllable current source or sensing components in the automation system. 161 School of Engineering Department of Electronic and Computer Engineering Equipment Automation for Remote Experimentation Supervisor: SHAO Qiming / ECE Student: ZAU Ka Ming / CPEG Course: UROP1100, Spring Since I am not at school, what I first did is to watch the video given by the professor. The video mainly talks about the analogue-to-digital converter (ADC) and the digital-to-analogue converter (DAC). For playing the music, it uses the DAC by transforming the electrical signal to analogue signal through loudspeakers. For talking on the phone, it uses the ADC by transforming the analogue signal from the microphone to the digital signal in the form of a radio wave emitting to the station. DAC and ADC are beneficial as there are many analogue things in nature such as temperature, sound, velocity. They are sensitive to noise and difficult to process and store analogue signal, but the digital signal can deal with these problems. However, during the conversion, some analogue signal information may lose because the analogue signal is continuous in time with the continuous amplitude. Therefore, one of the solutions is to control the sampling rate. Suppose the rate is very high; although it can get accurate data, the memory is not enough to support numerous data. If the rate is low, the data may inaccurate due to the aliasing effect. According to the Nyquist Sampling Theorem, the sampling frequency must be at least two times the maximum frequency of the input signal. Another solution is quantization, which assigns a sampled signal value from the discrete set of values. The resolution is how the assigned value close to the actual value is involved. Resolution is that full-scale range or reference voltage divide by two to the power of the number of bits. It means that if the range is extensive, the resolution is low. If the number of bits is large, the resolution is high. There is a tradeoff. To conclude what I have understood while watching the video. I understand the critical parameters of ADC and DAC are the resolution, reference voltage and sampling rate. Rational Design of Magnetic Devices Using Micromagnetic Simulations Supervisor: SHAO Qiming / ECE Student: FONG Cheuk Pan / PHYS Course: UROP3100, Fall Magnetic skyrmion, a type of quasi-particle, is interested by many electronic engineers. It is a great potential for futuristic electronic devices due to its high stability and small size. It can act as an information carrier which is transferred by energy-efficient spin current. Although the velocity and angle of motion are well investigated by many researchers, the deformation of skyrmion is still a difficult problem to understand. This problem leads to another important issue: recoverability of skyrmion size. Skyrmion expands due to magnetic field of materials and spin current. If it cannot return to size close to its initial state, skyrmion-based devices will not function properly, reducing the efficiency of devices. To understand the deformation and the recoverability of skyrmion size, this project is conducted. 162 School of Engineering Department of Electronic and Computer Engineering Rational Design of Magnetic Devices Using Micromagnetic Simulations Supervisor: SHAO Qiming / ECE Student: CHEN Zehan / PHYS Course: UROP2100, Spring Skyrmion is a topological object in ferromagnetic systems, promising for future high-density and non-volatile memory, where the skyrmion dynamics is a major concern. As a spintronic object, skyrmion can be driven by current-induced spin orbit torque (SOT), which will also cause the skyrmion to deform. This report will discuss the deformation of current-driven skyrmions using Thiele’s approach. Rational Design of Magnetic Devices Using Micromagnetic Simulations Supervisor: SHAO Qiming / ECE Student: XING Hexiang / ELEC Course: UROP1100, Summer Spin orbit torque (SOT), an emerging technology, improves magnetic random-access memory (MRAM)’s characteristics and makes it more competitive compared to other high-speed non-volatile memories (e.g., flash memory). It will accelerate the switching speed by highly reducing the incubation time. Among various spintronic devices, multilevel SOT device is strongly competitive in memory computing (AiMC) in deep neural network (DNN) area. Its intrinsic multilevel switching behaviour provides an efficient solution for matrix computing which is crucial in AiMC. For further detailed investigation in multilevel SOT devices, I run the micromagnetic simulation of multilevel MTJ based on Mumax3 programming. Front-End Mobile Application Development for a Locomotion Robot Supervisor: SHI Ling / ECE Student: FANG Yidong / ELEC Course: UROP1100, Spring With the advent of the age of technology, more and more household appliances can be controlled by electronic control systems. In this project, we aim to develop a front-end mobile application and its user interface for users to interact with an agile locomotion robot such as a floor sweeper. We just use MySQL to store the data and use Java to transfer the data to the cloud basement. From the hardware side, we initially planned to use ESP 32 to receive the data from the floor sweeper, however, for we need to control remotely, it is not enough to rely on WIFI to transfer data, we also need SIM 4G to help transfer data, so the WIFI version of ESP32 cannot complete our requirements, so we switched to the MKR1500 board which could use 4G to send the data. Finally, we obtain the map, send it to the cloud database and could display on Android based electrical devices. 163 School of Engineering Department of Electronic and Computer Engineering Multi-Agent System Control Supervisor: SHI Ling / ECE Student: XU Jinyun / ELEC FU Zhengyu / ISDN Course: UROP1100, Fall UROP1100, Fall A multi-agent system is a robotic system composed of multiple identical robots. It is commonly used to solve problems that are difficult or impossible for an individual agent to solve. In this research, we mainly focus on the formation control of multiple wheeled robots. We have conducted several experiments in system modeling, feedback control, and formation control of differential-drive robots. We are using the graph model to perform both distance-based formation and relative-state formation. Experiments are conducted simultaneously in the simulator and reality. Multi-Agent System Control Supervisor: SHI Ling / ECE Student: WU Bohuai / COSC Course: UROP1100, Spring Robotics is one of the most popular areas that many people would like to do research on this area. Therefore, to teach students about the essential and advanced technologies related to the robotics, there are many courses online that are relevant to robotics. However, the robotic courses online may not be useful enough for a student to know more about the robotics because using physical robot cars can give student more user experience. Thus, preparing more practical and interesting robotic courses and workshops is more important. This project sets up 7 lab materials for students to learn the useful techniques step by step. For all the 7 labs, we will use the robot car shown in Fig.1, which is a robot car with some LEDs and sensors. Multi-Agent System Control Supervisor: SHI Ling / ECE Student: ZHAO Xiaoqi / ELEC WANG Peiqi / ELEC Course: UROP1000, Summer UROP1000, Summer Multiple robot cars can perform different collective tasks, such as formation control and cargo delivery. The project aims at utilizing small two wheeled differential drive robots to accomplish tasks which would otherwise be expensive or even impossible for single machine. Robots used in the experiment run Raspberry Pi OS and communicate with the computer using WebSocket via Wi-Fi. Numerous position feedback information from the robot cars can be collected using methods like fiducial markers and motion capture systems. Various formation control algorithms have been tested such as virtual leader and potential field. A platform is also designed to automate the test procedures. Scenarios such as limited communication between each agent, limited position feedback and presence of obstacles may be simulated if such a platform could be developed. 164 School of Engineering Department of Electronic and Computer Engineering Development of Network Communication System for Locomotion Robots Supervisor: WONG Man / ECE Student: WANG Haoyi / SENG Course: UROP1000, Summer The aim of the project is to develop a network system for communication between autonomous robots and a cloud server, and among these robots. The most critical issue is to build a high-accuracy positioning system with an error of up to centimeters, which means that the accuracy of the GPS solution cannot meet the requirements. Here we build a localization system for robots. After comparing the pros and cons of various technologies, we chose to use the UWB (Ultra Wide Band) scheme for the localization system. We analyzed the communication algorithm of base stations and tags in the UWB module. Then, we build a prototype localization system. The system has four base stations to show the 3-dimensional coordinates of the tags. Two tags are put in the system to verify the stability, and the number of tags can be increased to 20. After doing experiments in different base station locations and heights, we achieved the expected accuracy in a small range of about 2 by 2 by 2 meters. The system also runs stably in a larger measuring area of about 20 by 20 by 2 meters, but the accuracy under that condition needs to be improved. 165 School of Engineering Department of Industrial Engineering and Decision Analytics Department of Industrial Engineering and Decision Analytics Big Data Analytics Techniques for Quality and Process Improvement Supervisor: TSUNG Fugee / IEDA Student: MU Yiteng / DA Course: UROP1100, Spring Traffic flow forecasting has been more and more critical nowadays. In this report, we give a literature review on current research around this topic. Based on what we have learned from the past experience, we aim to build a model based on the historical data from MTR and predict the traffic flow during the operation hours. We extract useful information from the raw data and use ARIMA (Autoregressive Integrated Moving Average) technique, which is a widely used model in this field. Moreover, we also tried different models for different types of stations to get the best performance of prediction. Finally, we did an evaluation of the current limitation and defects based on the forecasting results as well as the research progress. Data analysis in FinTech Supervisor: ZHANG Jiheng / IEDA Student: LIN Ruihan / DA WONG Siu Tim / DA Course: UROP1100, Summer UROP1100, Summer In this project, we aimed to develop an auto-trading strategy on cryptocurrency market using technical indicator combined with machine learning algorithms. We discovered two main technical indicator -- SAR & MACD to help catching the market opportunities, and used random forest classifier to reinforce the signal accuracy. By connecting to the server on IEDA department, we performed rigorous backtest and develop interactive plots for visualization. Lastly, we connect to the exchange by using external API -- CCXT to achieve real-time auto trading. Data analysis in FinTech Supervisor: ZHANG Jiheng / IEDA Student: XIAO Hanyu / SSCI Course: UROP1100, Summer In this report, we present some data analysis skills with python. We will first show how to use some python packages and library tools, such as pandas, regular expressions, scipy.stats, to process rough data and the process of getting CSV, HTML files, manipulate these data using regex (regular expressions). Next, through some useful data analytic tools in python, like aggregation, merge, group by, form a helpful data frame, and finally use scipy.stats tools to calculate the correlation index. Finally, we will also point out some flaws and limitations of this data visualization and put forward some proposed methods to improve it. 166 School of Engineering Division of Integrative Systems and Design Division of Integrative Systems and Design Synthetic Tactile Sensing For Prosthetic Fingers Supervisor: LEUNG Suk Wai Winnie / ISD Student: LIU Hanmo / CPEG Course: UROP1100, Summer It has been a long time that people imagine that artificial fingers can simulate the sensing function of real hands. By adding multiple sensors, the artificial fingers may accept the different environmental signals such as pressure, fracture, temperature, etc. Sensing the signals is the first step for prosthetic fingers. This research focuses on how to construct one appropriate circuit which can apply to specific sensors to fulfill the requirement of the gloves. A Multi-Source Semantic 3D Modeling Platform for Virtual HKUST Supervisor: YEUNG Sai Kit / ISD Student: PANG Ho Kiu Holdee / MAE Course: UROP1100, Spring The aim of this project is to create a system that can generate a 3D virtual model for HKUST using object segmentation. RGB-D sensor and other computer vision techniques to acquire the 3D virtual model of HKUST. For the current stage, we are tasked with two main duties, understanding the methods of image recognition and object segmentation, and designing a robot for travelling, detecting and manipulating objects. The final semantic 3D model will be useful in many ways, such as virtual walkthrough (virtual reality), robotics training, route and evaluation planning, campus re-design/remodeling. HKUST in VR Supervisor: YEUNG Sai Kit / ISD Student: LAI Kam Wai / ISDN TSE Yui Kai / ISDN Course: UROP1100, Summer UROP1100, Summer This project requires the use of Unity and VR equipment to create a virtual environment for laser cutter training. Tutorials on Unity VR are taken to familiarize with the application of the software. Training for manufacturing tools used to be tedious and time-consuming, since one-to-one training is usually required for students to be familiar with the operation, which takes technicians so much time and effort to repeatedly teach individual students. Hence, training using virtual reality might be a way to increase the efficiency of training since students can be trained without technicians and it does not require multiple physical devices for training. Currently, laser cutter, water-jet cutter, 3D printer, and welding machine in virtual space are built to make a virtual ISD Works! for students to interact with machine training. 167 School of Engineering Department of Mechanical and Aerospace Engineering Department of Mechanical and Aerospace Engineering Power-Source-Free Energy Generator Supervisor: CHEN Sherry / MAE Student: ZHANG Yuchen / AE Course: UROP3100, Fall This semester’s work involves the experimental preparations for the design of Energy Conversion Module (ECM). Two major steps were taken into consideration at the beginning: the first was to use a CAD product to simulate the realistic operation of a CPU chip, while the next was to use phase-transforming material to obtain the data of generated electricity under simulative conditions provided by the CAD product. Since there was shortage in phase-transforming material, I was asked by Dr. Chenbo to investigate the material composition and therefore fabricate the material that suits best to ECM. The latter procedure will hopefully be accomplished by next spring, after which all requisites will be attained, and a prototype can then be manufactured. Power-Source-Free Energy Generator Supervisor: CHEN Sherry / MAE Student: ZHANG Yuchen / AE Course: UROP4100, Spring The energy conversion module (abbreviated to ECM)’s operation notably depends on the pyroelectric material. To ensure standard functionality and further enhance its performance, different compositions of pyroelectric materials were fabricated to prepare for the assembly of the ECM during the semester. Being suggested several candidate compositions, the subsequent preparatory works focused on their realization using available ingredients and equipment. At the end of the term, there were in total two pyroelectric material rods ready for characterization. In addition to the successful attainment of the required materials, hands-on experiences of laboratory work, including safety issues and the use of various instruments, were obtained. Electrochemical Characterization and Evaluation of Thin Film Cathodes for Solid Oxide Fuel Cells Supervisor: CIUCCI Francesco / MAE Student: CHAWLA Anhad Singh / SUSEE Course: UROP1100, Spring This progress report discusses the results of using a new method Ultra-fast High-Temperature Sintering(UHS) on symmetric cells and protonic ceramic fuel cells. BaFe0.9Zr0.1O3-δ(BFZ) and BaFe0.9Zr0.1O2.9-δF0.1(BFZ-F) are compared as cathodes to gauge the performance of Fluorine as a dopant. Optimal conditions and experimental procedures for UHS are established to reach sintering temperatures for the cathode. Symmetric cell area specific resistance is calculated via electrochemical impedance spectroscopy(EIS) and compared to traditional, furnace-sintered material. Fuel cell performance is currently being evaluated and future research directions are listed. UHS performance and benefits are discussed and avenues of improvement are identified. 168 School of Engineering Department of Mechanical and Aerospace Engineering Electrochemical Characterization and Evaluation of Thin Film Cathodes for Solid Oxide Fuel Cells Supervisor: CIUCCI Francesco / MAE Student: CHAWLA Anhad Singh / SUSEE Course: UROP2100, Summer This progress report discusses the improvements achieved in the using of a new method Ultra-fast High- Temperature Sintering(UHS) on GDC symmetric cells. BaFe0.9Zr0.1O3-δ(BFZ) and BaFe0.9Zr0.1O2.9-δF0.1(BFZ-F) are used as cathodes and their UHS’d versions are compared to traditionally sintered BFZ. Optimal conditions achieved before are replicated with parameters that are more practical to the equipment used for UHS. Comparison of UHS and traditional furnace approaches to sintering the cathode are compared via Electrochemical Impedance Spectroscopy (EIR) to assess area specific resistance. Using the traditional method as a control reference, the UHS method shows definite improvement in lowered resistance, with Fluorine doped cathode showing even better performance. Fuel cell testing for other symmetric or half cells failed, preventing testing of UHS. Multidimensional Modeling of Nanoionics Supervisor: CIUCCI Francesco / MAE Student: PANG Ho Kiu Holdee / MAE Course: UROP1100, Fall Nanoparticle systems refers to systems that are made of nanoparticles, size ranging from 100nm to 500nm. In the past few decades, these systems are being utilized in various applications, including fuel cells, semiconductors and medical drug delivery, to name but a few, due to their unique properties. Nanoparticle have unique optical, electrostatic and mechanical properties that propelled the development of various new technologies in recent years. This project will be focusing on the electrostatic properties of nanoparticles and their impacts on solid fuel cells. To start with, a brief introduction of the principles of solid fuel cells will help us identify with the electrostatic responses of nanoparticles. A fuel cell1 is a device that utilizes chemical redox reaction to generate electricity, where main components includes a positive electrode called cathode and a negative electrode called anode and a solution called electrolyte, providing a medium for ionic exchange between the electrodes. Reduction reaction occurs in the cathode and oxidation rection occurs in the anode, which the combined effect causing the electrons to be utilized into generate a electric current, for example this is the basic reaction2 of a hydrogen fuel cell: 2H2 + O2 →2H2O + 2e-. Now it can be seen why nanoionic materials are employed in battery technologies, the large exposed surface area of nanoparticles making their electrostatic interactions readily adjustable, by controlling factors like concentration of ions present in solution and surface charges. 169 School of Engineering Department of Mechanical and Aerospace Engineering Polymer Composite Batteries Supervisor: CIUCCI Francesco / MAE Student: YU Tsz Tat / MAE Course: UROP4100, Fall Reducing the thickness of lab-made solid-state polymer electrolyte remains an engineering challenge because of the low mechanical strength of the common polymer materials. Thus, a novel polymer electrolyte was fabricated using thin, porous, and lightweight polytetrafluoroethylene (PTFE) as a mechanical backbone, with infiltration of a dual salt polymer electrolyte using lithium bis(trifluoromethane- sulfonyl)imide (LiTFSI), lithium bis(oxalate)borate (LiBOB), poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP), fluoroethylene carbonate (FEC) and adiponitrile (ADN). The resultant membrane delivered low resistance (2.2 Ω) and a wide electrochemical window (5.0V). The electrolyte was coupled with lithium metal anode and nickel-manganese-cobalt cathode, and the batteries operated at 2 C for over 500 cycles with high capacity (112 mAh/g), high capacity retention (89%), and high coulombic efficiency (99.776%). Polymer Composite Batteries Supervisor: CIUCCI Francesco / MAE Student: YU Tsz Tat / MAE Course: UROP4100, Spring Lithium-ion batteries, especially solid-state batteries, hold great promise in sustainable energy. However, the narrow operational temperature window (-20 ℃ to +55 ℃) is a major problem yet to be resolved, limiting the practical applications of lithium-ion batteries. In this work, an in-situ polymerized gel polymer electrolyte using 1,3-dioxolane (DOL) and LiPF6, with methyl propionate (MP), fluoroethylene carbonate (FEC), and LiTFSI was developed. The resultant gel polymer electrolyte delivered exceptional ionic conductivity of 3.43 mS/cm2 at 20 ℃ and 1.53 mS/cm2 at -30 ℃. The Li|LiFePO4 cell displayed a high specific capacity of 146.7 mAh/g and coulombic efficiency of 99.93%. Moreover, the cell delivered outstanding thermal stability and was capable of operating with long-term stability at -20 °C. Extended Ortho-K Lens for Myopics Supervisor: LAM David Chuen Chun / MAE Student: SEO Minji / AE Course: UROP1100, Spring UROP1000, Summer Over the past decades, the high prevalence of myopia has been recognized especially in Asian countries and this has encouraged ophthalmologists to investigate a feasible solution to reduce the myopia progression. This report proposes a possible means to apply a temporary change in corneal shape while it discusses the overall progress in the project. By specifying the limitations on the initial experimental setup, tools and controlled variables, this report shows the improvement made on the aforementioned factors and their significances. In the coming summer, it is planned to start the data collection for porcine eyes and the animal handling practice for rabbit eyes. 170 School of Engineering Department of Mechanical and Aerospace Engineering Smart Contact Lens Supervisor: LAM David Chuen Chun / MAE Student: POYYAMOZHI Sricharan / CBME Course: UROP1100, Fall This paper presents a methodology to analyze the chemical solutions used during PCB fabrication after photolithography. The main baths, developing solution, etch bath and strip bath along with their rinse cycles are isolated and analyzed using Ultra-Violet Spectroscopy. The finished coils are also checked for defects using Scanning Electron Microscopy. The data collected with these methods are processed using matlab to give simple contamination and quality gauges. The UV-Spectroscopy was found to be exemplary other than when analyzing etching solution which has transition metal ions resonating across a wide spectrum. SEM was found to be a capable identifier for streaking but inefficient for whisker identification. Smart Contact Lens Supervisor: LAM David Chuen Chun / MAE Student: POYYAMOZHI Sricharan / CBME Course: UROP2100, Spring In this paper major defects that are common during Printed Circuit Board Manufacturing and their causes are introduced. Process optimization during initial fabrication during photolithography and photoresist development are considered to reduce error. The defects, with respect to specific chemical baths are considered and monitoring protocols are established to identify dirtiness thresholds. The monitoring of etching baths are fully determined. Whisker formation is characterized to additive PCB fabrication and a basis for future electroplating experiments is established. A new circuit design is introduced to uniquely induce and research whiskers by concentrating electricity at sharp turns and research the effects of scale on whisker formation. Application of AI-based Technique to Enhance Thermal Comfort Sensing for Smart Air Conditioner Supervisor: LEE Yi-Kuen / MAE Student: ZHENG Tianshi / COGBM Course: UROP1100, Spring In this project, we need to make improvements based on the previous system, which combines the Bluetooth connection with Arduino and Web database sending using Android App. The app collects data from the mobile phone’s built-in sensors, and receives other parameters through Bluetooth from Arduino MCU, finally compute the output of the PMV value. During this semester, I developed an iOS app that has the similar functions to the previous Android one using XCode and Swift language. 171 School of Engineering Department of Mechanical and Aerospace Engineering Application of AI-based Technique to Enhance Thermal Comfort Sensing for Smart Air Conditioner Supervisor: LEE Yi-Kuen / MAE Student: CHOW Hong Kiu / IIM Course: UROP1000, Summer This project aims to develop a low-cost system detecting thermal comfort based on the predicted mean vote(PMV) and predicted percentage of dissatisfied (PPD) with application of Arduino, MIT App Inventor and sensors including temperature and humidity sensor, air velocity sensor and motion sensor. Due to limitation of current PMV model, which is low accuracy, an improvement on the current model is worth considering. A survey is conducted to compare the difference between PMV and PPD based on current equation and observed mean vote (OMV) and observed percentage of unacceptability (OPU) to access the accuracy of the model and factor affecting the accuracy. I am involved in the Arduino coding and connection to sensors, MIT App Inventor on IOS device and the conduction of survey. Application of AI-based Technique to Enhance Thermal Comfort Sensing for Smart Air Conditioner Supervisor: LEE Yi-Kuen / MAE Student: LIM Her Wei / SENG Course: UROP1000, Summer Heating, Ventilating, and Air-Conditioning Systems (HVAC) use 24% of energy consumption in the commercial sector, according to an EMSD research article (CLP Power, 2021). However, these systems are not very efficient, causing a waste in both energy and lower thermal comfort, which is the main goal to improve in this research. To better the method of controlling HVACs will be crucial for reducing energy consumption and improving the comfort level of users. With this said, a smartphone app can be developed for the calculation and surveying of PMV, facilitating the process and increasing efficiency, while making a foundation for further smart-HVAC projects. Application of AI-based Technique to Enhance Thermal Comfort Sensing for Smart Air Conditioner Supervisor: LEE Yi-Kuen / MAE Student: SALIM Richie Wee / MATH-CS Course: UROP1000, Summer This paper will report the result of a data model created from machine learning methods using human pose data that has been collected before. Machine learning is used to compare the conventional models of equations and get the accuracy of the new model given the data. This is a crucial step in determining the user’s metabolism rate based on their current activity, so we can later determine the PMV for further processing. The model inputs our three-dimensional position processes it and classifies it into one of the basic human activities e.g. sitting, walking, running, or standing. Based on a widely used machine learning platform, the model is optimized for classification and reduces losses during training, resulting in a good performance and data accuracy. 172 School of Engineering Department of Mechanical and Aerospace Engineering Application of AI-based Technique to Enhance Thermal Comfort Sensing for Smart Air Conditioner Supervisor: LEE Yi-Kuen / MAE Student: MO Yin Chung / MAE Course: UROP1100, Summer With the increasing demand for smart buildings, professionals are looking into indoor thermal comfort to cater to people’s needs and consider the energy consumption of buildings. This project focuses on utilizing MEMS sensors to conduct thermal comfort measurements (humidity, air velocity and temperature) in buildings and a smartphone app that consider human factors and activities to evaluate the metabolic rate. By obtaining these data and doing a massive number of surveys, we can determine the Predict Mean Vote (PMV) index and use it to compare with the peoples’ subjective feelings. Moreover, it can also improve the accuracy of Human Thermal Comfort Sensing systems and achieve a smarter Heat Ventilation and Air-Conditioning system. Application of Artificial Intelligence to Enhance the Fluorescence Microscopy of Circulation Tumor Cells Captured by MEF Chips Supervisor: LEE Yi-Kuen / MAE Student: TANUWIJAYA Richard Valent / SENG Course: UROP1000, Summer Circulating tumour cells (CTCs) in the blood sample can be very useful to identify the current patient’s needs and medications. However, detection and identification of CTCs can be time consuming if done by manual operations. Therefore, a machine to automate the fluorescence microscopy of the captured cancer cells and CTCs is invented. Besides, a program to automate the identification process is needed. Thus, in this report, we will discuss how to detect potential circulating tumour cells (CTCs) in the blood samples with the help of fluorescence microscopy and digital image processing. This report includes the characteristics of circulating tumour cells and white blood cells and explanation of ImageJ macro language program to identify potential CTCs. Application of Artificial Intelligence to Thermal Comfort Sensor App for New-Generation Smart HVAC Systems Supervisor: LEE Yi-Kuen / MAE Student: HUNG Ling Hon / COSC Course: UROP1100, Spring The majority of people today spend most of their time indoors, and the indoor environment can be closely related the people's working efficiency. The Heating Ventilation and AirConditioning (HVAC) systems in buildings provide proper indoor air quality and thermal comfort to people inside. To better provide a thermally comfortable environment, the system needs to refer to a quantitative measure. Predicted Mean Index (PMV) is one of the most commonly used measures of thermal comfort. This research project aims to continuously improve the PMV measuring system proposed by Izhar et al. I have explored an alternative way to get environmental data, explored methods to program and debug a sensor used in the project, and collected more data for one of the algorithms determining people’s metabolic rate. 173 School of Engineering Department of Mechanical and Aerospace Engineering Development of AI-based Smartphone Microscopy system for Identification of Algal species for Environmental Protection Supervisor: LEE Yi-Kuen / MAE Student: THIESMAN Wilson / SENG Course: UROP1000, Summer Algal blooms have been a critical problem to the marine environment. It poses significant challenges to water security, fishing management, and the food-chain of the marine ecosystem. Although several protocols are available to tackle this problem, they exhibit high amount of cost which prevent us to increase the accessibility of using these protocols, especially in rural areas. This problem indicates that a cheaper and more efficient method is required to solve this problem even further. This report focuses on developing a low-cost microscope to identify algal species based on their characteristics. Furthermore, we utilize the Raspberry Pi computer and python as the fundamental of our project and use an autofocusing camera manufactured by ArduCam Company to enhance the performance of our microscope. Development of low-cost rapid pathogen detection microsystem in COVID-19 Pandemic: Micro PCR module- Bioengineering Approach Supervisor: LEE Yi-Kuen / MAE Student: FERNANDO Lindamulage Yasiru Hansaja / SENG Course: UROP1000, Summer With the COVID-19 pandemic still at large, with over 210 million confirmed cases worldwide and 4.4 million deaths as of August 2021, rapid and constant testing has become vital at stopping the spread of this pandemic and future pandemics. Thereby, it is vitally important to identify new methodologies by which we can reduce the cost per test and the testing time. This is where Electro-Gated FET Biosensors (EGFET), which are an iteration of the popular ISFET but with a discrete setup of the gate material in such a way that it can be replaced after its lifetime, comes into play as they provide a cheaper and faster alternative to the detection of target DNA strands. The primary focus of this report is to understand the structure and process that facilitates the usage of EGFETs and EIS in DNA detection and the relevant factors in optimizing this sensor. Development of low-cost rapid pathogen detection microsystem in COVID-19 Pandemic: Micro PCR module- Bioengineering Approach Supervisor: LEE Yi-Kuen / MAE Student: LAU Siu Man / AE Course: UROP1000, Summer For detecting the COVID-19 pathogens, only little amount of DNA is present in the sample. It is not enough for undergoing testing, so a duplication of DNA amount is required. In medical firms, PCR is done in order to increase the amount of DNA to be tested. PCR requires heating of sample to certain temperature and stay for certain period of time, then cool down to another temperature and remain certain period of time. In this UROP period, I have learnt to use COMSOL for different stimulation. This report will talk about the final stimulation done for heating up some sample using a kit provided by the team. 174 School of Engineering Department of Mechanical and Aerospace Engineering Development of low-cost rapid pathogen detection microsystem in COVID-19 Pandemic: Micro PCR module- Bioengineering Approach Supervisor: LEE Yi-Kuen / MAE Student: LEUNG Kit Hin / AERO Course: UROP1000, Summer This report discusses the finding of the heating experiment of a MCH heating board with the use of Arduino kit and code. After conducting the heating experiment, it was found that temperature of the heating board increased in a faster manner when the input voltage was higher and shorter time was needed for it to attain the required temperature. Apart from the above finding, there was another observation made in the experiment. In order to heat up the board to a higher temperature, a larger input voltage is required to complete this task because of the different temperature limits for various voltage values. It might take a long time for the heater to attain a higher temperature if the voltage is kept constant. Development of Low-cost Smartphone-based Microscope for Global Health Supervisor: LEE Yi-Kuen / MAE Student: LEUNG Chi San Ivor / PHYS Course: UROP1100, Summer The first prototype of a low-cost fluorescence microscope is constructed. The prototype comprises a credit-card-size Raspberry Pi computer integrated with a high-quality camera module, commercial microscope objective lenses and other materials such as 3Dprinted components. Its performance is evaluated and compared with a commercial microscope, Olympus IX 70, in the bright-field mode in the following aspects: resolving power, image quality, and signal-to-noise ratio. Assessment results show that the prototype performs comparably to the commercial microscope. In particular, it demonstrates a higher signal-to-noise ratio than the commercial microscope. Besides the performance analysis, we also discuss some possible improvements to turn the prototype into a more suitable equipment in low-cost point-of-care medical diagnostic systems. 175 School of Engineering Department of Mechanical and Aerospace Engineering

Ambular: An Emergency Response Drone Supervisor: LI Larry / MAE Student: AYDOGMUS Emre Can / MAE KAN Ho Hung / AE SRIBHIBHADH Runn / SENG SWAMINATHAN Srivatsav / AE Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer UROP1100, Summer Ambular, an aviation project initiated by the International Civil Aviation Organisation (ICAO), aims to develop electric Vertical Take-off and Landing (eVTOL) technology for improving people’s quality of life. Initial testing of the current Ambular design shows several aerodynamic issues, the biggest being the negative lift generated by the fuselage and propeller arms. Thus, the current stage of the project focuses on redesigning the fuselage and optimizing the arm shape. First, preliminary research on the different fuselage and airfoil designs were conducted. The information gathered was then used to guide the development of new Ambular components. A few designs were considered for both the fuselage and propeller arms, which were then tested via computational fluid dynamics (CFD) software to determine one best suited for the mission parameters of low-speed flight and long endurance. Development of Extended-Range Unmanned Surface Vehicles (USV) Supervisor: LI Larry / MAE Student: CHONG Long Fung / MAE Course: UROP1100, Spring The potential energy sources for Unmanned Surface Vehicles (USVs) have been investigated. Methods of harvesting energy and its mechanisms are represented in the paper. Wind energy, wave energy and solar energy are the main topics of this research, and these energy mechanisms are analyzed and their performance benchmarked against each other. 176 School of Engineering Department of Mechanical and Aerospace Engineering Early Detection of Thermoacoustic Instability in Solid Rocket Motors Supervisor: LI Larry / MAE Student: CHAN Yan Yan / PHYS-IRE Course: UROP1100, Spring In this project, the nonlinear transition from low-amplitude aperiodic pressure fluctuations (i.e. combustion noise) to high-amplitude limit cycle oscillations inside a full-scale solid rocket motor (SRM) is studied using multifractal detrended fluctuation analysis (MFDFA) and spectral entropy ES of time series data. It is shown that the noise-like pressure fluctuations observed before combustion instability is scale-invariant and features a multifractal structure that then disappears at the onset of thermoacoustic oscillations. The fractal structures of the combustion noise and the thermoacoustic oscillations are quantified by the generalised Hurst exponent Hq and the multifractal spectrum. Additionally, the variation of the Hurst exponent H and Es as a function of time is examined, and the feasibility of using H and Es as early warning signals of an impending thermoacoustic instability in a full-scale SRM is demonstrated. Early Detection of Thermoacoustic Instability in Solid Rocket Motors Supervisor: LI Larry / MAE Student: GANI Bryan Suryaraso / MAE Course: UROP1100, Summer Thermoacoustic instability may result in excessive vibration and heat transfer in a combustion system. However, the nonlinear interactions occurring between the flow, acoustic and chemical processes make it difficult to control the system using a linear approach. Nowadays, researchers are trying to implement machine learning as a tool to eliminate or suppress this instability. This project reviews some of the latest progress made by researchers in the field. According to the literature, there are two primary research approaches. The first uses data from pressure and heat-release rate measurements. The second uses imaging and deep learning. It is acknowledged that some of the topics might need further study if they are to be deployed for the suppression of thermoacoustic instability. Nonlinear Dynamics and Closed-Loop Control of Droplet Formation in a Capillary Jet Supervisor: LI Larry / MAE Student: OLSSON Christopher Sven / AE Course: UROP3100, Fall A brief review of the currently available literature on propeller tip design for micro air vehicles is presented. There is currently no universal stand-alone metric for quantifying the efficiency of a propeller tip (energy and thrust lost). At low Reynolds numbers (Re < 50,000), propeller tip effects become increasingly important, and this effect is explored in detail with reference to potential applications in micro air vehicles. 177 School of Engineering Department of Mechanical and Aerospace Engineering Aircraft Design and Air Transportation Operations: How do they Affect Each Other? Supervisor: LIEM Rhea Patricia / MAE Student: CHAVANACHINDA Phisitchai / MEGBM Course: UROP1100, Fall With the rising demand for air travel, airports around the world find themselves becoming increasingly busy. The traffic congestion and high demand has propelled airlines to explore the use of larger airliners. This paper aims to analyze different ranges of flights and their respective fuel efficiencies, to examine the correlation between payload and efficiency, and to identify the most suitable scenarios for long range aircraft, specifically the B777-300ER. The study helps develop a deeper understanding of the linkage between aircraft design and air transportation operations. The results show that short range flights have higher variance and fluctuations, affecting the accuracy of fuel consumption predictions. Moreover, the results illustrate that the most efficient route distance for long haul aircraft is the medium range, contradicting presumptions about long range aircraft while substantiating the rationale behind utilizing large aircraft on shorter routes. Aircraft Design and Air Transportation Operations: How do they Affect Each Other? Supervisor: LIEM Rhea Patricia / MAE Student: SWAMINATHAN Srivatsav / AE Course: UROP1100, Fall UROP2100, Spring With the rising demand for air travel, airports around the world find themselves becoming increasingly busy. The high passenger demand across different flight routes have incentivised certain airlines to consider using long-range aircrafts such as the Boeing 777-300ER for all variants of flight routes. This paper aims to analyse the fuel burn-offs of the Boeing 777-300ER across different flight routes considering the distance travelled and net flight time. The study helps in developing a deeper understanding of the linkage between aircraft design and air transportation operations. The results show that short-range flight routes have a greater divergence of fuel burn-offs primarily due to the high payload variations across a relatively smaller distance. Furthermore, the results also depict the existence of a critical passenger payload at which the aircraft performs the most optimally. Aircraft Design and Air Transportation Operations: How do they Affect Each Other? Supervisor: LIEM Rhea Patricia / MAE Student: YIU Kwan Ho / AE Course: UROP1100, Fall Conditions in Top Level Requirement is considered as the most efficient way in aircraft operation. However, it is hard to operate aircraft only in such specific conditions. To optimize fuel efficiency, the optimum operation pattern of the aircraft needs to be found. Taking B777-300ER as an example, the longer flight distance, longer flight time and higher payload boost the fuel efficiency. Attaining these conditions, the aircraft tends to operate more fuel efficiently. 178 School of Engineering Department of Mechanical and Aerospace Engineering Aircraft Design and Air Transportation Operations: How do they Affect Each Other? Supervisor: LIEM Rhea Patricia / MAE Student: CHAVANACHINDA Phisitchai / MEGBM Course: UROP2100, Spring With the increasing range of newer aircraft, flights are becoming longer and farther than ever before. It is a common practice for pilots to plan step climbs in long-haul flights in order to maximize the fuel efficiencies and the take-off weight. This study aims to analyze different flight mission profiles of short, medium, and long-haul flights, to examine the correlation between the number of step climbs, the take-off weight, and the respective fuel efficiencies, and to identify the suitable flight trajectory and planning. The results show that increasing the number of step climbs on long-haul flights can improve the fuel efficiency during the cruise segment. This paper helps to develop an understanding of the links between aircraft design and air transportation operations. Aircraft Design and Air Transportation Operations: How do they Affect Each Other? Supervisor: LIEM Rhea Patricia / MAE Student: YIU Kwan Ho / AE Course: UROP2100, Spring For economical and environmental reasons, the aviation industry urges the even more fuel-efficient aircraft operation. Studying about how operational factors affect fuel efficiency, the optimum conditions for air transport can be predicted. This study focuses on two operational factors only, the number of step climb and TOW. The research results find the more step climb ensures aircraft operate at the optimum altitude and the higher TOW results in lower fuel burnoff per weight. For the sake of fuel efficiency improvement, aircraft should perform frequent step climb to reach the optimum altitude and carry the higher payload. Study of Flapping Fins of a Bio-inspired Manta Ray Robot Supervisor: QIU Huihe / MAE Student: HO Ka Ying / MAE Course: UROP1100, Fall The flapping motion of the manta ray has always been a research interest since manta ray is considered to swim with high efficiency. Its efficiency is largely related to it’s the flexibility of its fin and its swimming mechanism involves twisting in both spanwise and chord wise direction, and it is also related to wake resonant theory, which is very different from the mechanism we have for current underwater vehicle. Many studies have been done to unveil manta ray’s swimming secret. This report will give a brief summary of the kinematics of manta ray’s flapping motion (dorsoventral oscillation), introduce some inspired designs of manta ray robots and some thoughts for future study. 179 School of Engineering Department of Mechanical and Aerospace Engineering Study of Flapping Fins of a Bio-inspired Manta Ray Robot Supervisor: QIU Huihe / MAE Student: HO Ka Ying / MAE Course: UROP2100, Spring To achieve high swimming efficiency, a bio-inspired manta ray robot must be flexible to replicate the delicate flapping motion of a manta ray. However, the flapping mechanism should remain simple as the robot should be compact for practical application. In contrary to conventional “bar and linkage” design, soft robotic mechanism may offer simple solutions in replicating flexible motion. In this report, a soft robotic mechanism, fin ray structure will be explored. An experiment analyzing the flapping performance of a robotic fin will also be discussed, with details about the experimental setup and methodology (Particle image velocimetry). Possible future approaches will be proposed based on the discussions. Environmental Impact by Air Traffic: Assessing Aircraft Noise nearby HK Airport Supervisor: REDONNET Stephane / MAE Co-supervisor: HORNER, Andrew Brian / CSE Student: CHAN Ngai Nam / AE Course: UROP1100, Summer The Aircraft Noise Assessment project focussed on obtaining on-site aircraft noise data to provide verifications for an existing aircraft noise level prediction software developed by HKUST Aerospace researchers. The project was planned and executed in three consecutive phases of preparation, on-site data acquisition and signal post-processing, relying on audio recording equipment and data processing software to conduct the required research, with development of an aircraft noise data Test Suite. The results of the signal processed were compared to the prediction software to conduction error analysis and initiate possible modifications and improvements, and the Test Suite could be continuously utilised in parallel to validate the prediction software in future. Experimental Characterization of a Modified Airfoil Offering Higher Aerodynamic Performances Supervisor: REDONNET Stephane / MAE Student: LEUNG Tsz Yin / AERO Course: UROP1000, Summer A new class of aircraft known as miniature unmanned aerial vehicles is becoming popular for everyday tasks. These light-weight aircraft operate within low to moderate Reynolds numbers and have very simple high-lift devices. An unconventional airfoil with slits added across their upper and lower surface promises to offer improved performance at high angles of attack at Re 6 million was tested at Re 133 000. This study confirmed the findings of a previous study and found the optimal slit location is independent of the Reynolds number, while the thickness of the slit should be increased as the Reynolds number decreases. 180 School of Engineering Department of Mechanical and Aerospace Engineering Robotics Gripper Development for Warehouse Automation Supervisor: WANG Michael Yu / MAE Student: MA Ho Yeung / MAE Course: UROP1100, Spring To automate the warehouse object picking and sorting process, a gripper that capable to handle objects with different shapes and textures is preferred as the shapes and textures of the objects in warehouses may not be consistent. However, most of the robotics grippers nowadays are targeted to a specific shape and texture. The research team is aimed to design a new universal gripper that capable to handle objects with a wide range of shapes and textures and the fabrication method of it. To achieve that, a novel technology, gecko-inspired adhesion, is applied to the design of the gripper. The targets of the project are to design an experiment equipment for the research team to experiment with the performance of the self-made gecko-inspired adhesive sticker and design a mold for fabricating a suspension structure for the gecko-inspired adhesive sticker. Robotics Gripper Development for Warehouse Automation Supervisor: WANG Michael Yu / MAE Student: LEE Man Yeung / MAE Course: UROP2100, Summer To facilitate the manual transportation of small delicate objects like wafers and mobile phone glass monitors, an handheld gripper was purposed to develop to replace existing vacuum grippers. This handheld gripper makes use gecko-inspired dry adhesive material to realise agile attachment and detachment mechanism. Different prototypes were fabricated to acheieve ergonomic features and quick replacement mechanism of the adhesive material. The prototypes introduced in this report use magnets to act as energy storing structure and have different quick replacement mechanism. The performance of these prorotypes are tested, and the problems and solutions are suggested. The experience learnt from testing these prototypes can be referenced by the future designs. Digitalization of Wet Lab Project Supervisor: YANG Jinglei / MAE Student: DU Xiawei / CPEG PANG Chun Lung / MAE WANG Yucheng / SENG ZHANG Rushan / SENG Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer UROP1000, Summer We worked with the WeShare Technology team this summer. Safety risks in the current chemistry laboratory have been identified through the background investigation and two field trips. A physical X-Tray system (the name of our hardware device) is designed and integrated with back-end programs and an interactable user interface. It is used to track the placement and usage of chemicals in the laboratory to lower the risk of safety accidents due to improper storage and accidental reactions between dangerous chemicals. A chemical replenishment notification function was also added to the system to enhance the convenience of laboratory users. This report is going to detailly introduce the logic and function design behind our software part and our hardware part. 181 School of Engineering Department of Mechanical and Aerospace Engineering EndoRob Project Supervisor: YANG Jinglei / MAE Student: CHAN Cheuk Yi / SENG Course: UROP1000, Summer Different types of medical imaging techniques are important for understanding the human digestive system and making appropriate diagnosis to patients. In order to examine the inner surface and cavity of hollow organs along the gastrointestinal track, invasive imaging techniques are used. However, these techniques have their limitations including limited vision and causing discomfort to patients. This report analyzes current invasive imaging products and their limitations, then propose a plan to design a mini robot with inchworm-like movement that can walk through the gastrointestinal tract to take information. Processing and Analysis of Smart Coatings on Campus Buildings: Part I Supervisor: YANG Jinglei / MAE Student: SEO Minji / AE Course: UROP1100, Fall Different types of medical imaging techniques are important for understanding the human digestive system and making appropriate diagnosis to patients. In order to examine the inner surface and cavity of hollow organs along the gastrointestinal track, invasive imaging techniques are used. However, these techniques have their limitations including limited vision and causing discomfort to patients. This report analyzes current invasive imaging products and their limitations, then propose a plan to design a mini robot with inchworm-like movement that can walk through the gastrointestinal tract to take information. Windows Cleaning Robot Supervisor: YANG Jinglei / MAE Student: LAU Yuen Hei / SENG Course: UROP1000, Summer Recently, the domestic service robot market including window cleaning robot and vacuum cleaning robot is growing rapidly. Although their demand are keep growing, window cleaning robot is not a popular product due to the price and safety concerns. The aim of this study is to develop a domestic window cleaning robot that is cheap and robust. By analyzing the current window cleaning robots in the market, their pros and cons are found. Then, a robust robot was designed, and improvements are made while keeping the cost as low as possible. The new design tackled problems that people living in the city would face which could potentially open more demand for the industry. 182 School of Engineering Department of Mechanical and Aerospace Engineering 183 Radiative Transport in Transparent Aerogels for Solar Thermal Energy Applications Supervisor: YAO Shuhuai / MAE Student: YIP Chun Yin / MAE Course: UROP1100, Fall Solar spectrum provides unlimited potential in our life, and as building nowadays occupies more than 40% of energy nowadays, making updates to the building windows become an effective solution. Making the window itself become a supplementary energy source by absorbing the solar spectrum selectively, the problems of energy overconsuming will be solved. In this report, I, as a UG student, is helping HKUST professors to design and build this system up. There are two models currently constructed, one is from me and one is from a research associate professor, Li Weihong, who teach me a lot. In this report, each model’s features and significance will be discussed, and a comparison will be made.

Abstracts of UROP Projects 2020-21 School of Business and Management School of Business and Management Department of Accounting Department of Accounting Machine Learning and Sentiment Classification in Chinese Financial Text Supervisor: HUANG Allen Hao / ACCT Co-supervisor: YOU Haifeng / ACCT Student: HUNG Chak Lung / QFIN Course: UROP1000, Summer The market news and trend could usually affect the sentiment of both financial institutions and individual investors, thus changing their investment decisions and affecting the stock performance. In recent years, thanks to the technological advancements in computer science research, there is a breakthrough in natural language processing, and we can use different NLP models for analyzing subjective information and predicting the sentiment of readers. In this research, the team1 has collected and labelled over 14,000 and 15,000 sentences from analyst reports and company annual reports from SHH and SHZ. After labelling those sentences, I have fine-tuned a pre-trained NLP model – Chinses Bert model for sentiment analysis and predicting readers emotions. Among different fine-tuned Bert models, it is found that the BERT model finetuned by the 3 types of the labelled sentences has the best statistical performance. Machine learning and Sentiment Classification in Chinese Financial Text Supervisor: HUANG Allen Hao / ACCT Co-supervisor: YOU Haifeng / ACCT Student: WANG Meihan / RMBI Course: UROP1000, Summer This project aims to build a sentiment classification model for Chinese financial text in the context of insufficient Chinese financial NLP models. After the preparation which Includes text corpus collection and manual labeling, we chose 3 models to train and compare: Logistic Regression, Naïve Bayes, and Bert (including a base Bert model and a financial Bert model). We collected two sets of corpora for training: one is from the ‘Operation Analysis’ session of different corporate annual reports and the other one is from the analyst reports. The fine-tuned Bert model significantly outperforms others on both annual reports and analyst reports for the three-classification task. The codes and related documents can be found at: https://github.com/mwangbk/UROP2021 Machine learning and Sentiment Classification in Chinese Financial Text Supervisor: HUANG Allen Hao / ACCT Co-supervisor: YOU Haifeng / ACCT Student: LI Xiangyu / ECOF Course: UROP1100, Summer This study investigates the accuracy of the Bidirectional Encoder Representations from Transformers (BERT) model in classifying the sentiment of Chinese financial text. We first label 20,000 sentences from annual reports and analyst reports as positive, negative, or neutral, and then use them to fine-tune and test the Google Chinese BERT. Out-of-sample results suggest that BERT consistently outperforms benchmark bag-of-words methods, with higher classification accuracy and better cross-sample generalizability. This performance is stable across years, yet fluctuates across industries. Alternative pretrained BERT models yield qualitatively similar results. The overall findings suggest using BERT to build sentiment factors and guide equity investment. 186 School of Business and Management Department of Accounting Machine learning and Sentiment Classification in Chinese Financial Text Supervisor: HUANG Allen Hao / ACCT Co-supervisor: YOU Haifeng / ACCT Student: SHIU Heng / QFIN Course: UROP1100, Summer Bidirectional Encoder Representations from Transformers, or BERT, is a transformer-based machine learning model developed by google that has significantly changed the landscape of NLP. Compared to previous models that read texts sequentially, BERT’s key innovation is that it uses bidirectional learning to gain the context of a word based on the entire sentence. Furthermore, BERT is pre-trained on large amounts of data. For instance, the original BERT model released by Google is pre-trained on a large corpus that includes all 2.5 billion words from Wikipedia. Value Simplex FinBERT, a BERT model designed for Chinese financial text, is pre-trained with over 4 million finance-related articles. BERT’s extensive training data allows it to perform well even when applied on relatively small datasets. BERT is also designed to be highly versatile, allowing users to fine-tune its model for different tasks by simply adding a few output layers. Downstream tasks that BERT could accomplish include Extraction-based Question Answering, Natural Language Inference, Sentiment Analysis, and others. Machine learning and Sentiment Classification in Chinese Financial Text Supervisor: HUANG Allen Hao / ACCT Co-supervisor: YOU Haifeng / ACCT Student: TU Bingying / DSCT Course: UROP1100, Summer The goal of this research is to use the chinese BERT model to perform sentiment classification on annual reports from the companies used to compute CSI300. The first stage of this research was to extract sentences from annual reports and prepare them for training use. The criterion was set to label the sentiment of each sentence and the basic information of the sentences are collected. After preparing the data, chinese BERT model is utilized to perform sentiment classification. We tried to use the model to analyze the sentiment of annual reports and then facilitate the investors’ decision making. Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: CHAN Chak Him / ECOF LU Xiaoyi / ECON ZOU Haoxiang / RMBI Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall The commodity futures market in China has its own uniqueness in terms of the regulations and style of transaction, while some important indicates of the market (e.g. hedging pressure, liquidity, momentum, term structure, etc.) might perform differently from markets in the U.S. and in Europe. While the Chinese commodity futures market remains immature in terms of regulations and the behavior of inexperienced individual investors, few investigations have been conducted. This report aims to better understand the commodity futures in China by developing features and using them to design a portfolio trading strategy. At the same time, using statistical and machine learning methods to test the effectiveness and the consistency of the strategy, as well as examining the features’ impact on future returns. 187 School of Business and Management Department of Accounting Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: SHIH Jackson / QFIN SO Man Kit / QFIN TAM Shing Hang Boris / COGBM Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall Analyzing investment decisions with machine learning techniques is getting increasingly popular. In this report, we explored some simple machine learning models which make use of technical indicators to try to predict future market return in the short term. Our testing and training data are taken from over 1000 A-share stocks in around a 4 year timespan (from the beginning of 2000 to December 2003). We have chosen 8 useful technical indicators by looking at their t-statistics and used these indicators to fit with the Decision Tree and Logistic Regression Model. The 2 models take the indicators’ value and output binary signal on buying or selling. Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: SUN Beini / RMBI ZHANG Yichen / RMBI YIU Kwong Chun / QFIN Course: UROP1100, Fall UROP1100, Fall UROP2100, Spring UROP1100, Fall This paper has referred to Bartram and Grinblatt (2018) and tried to estimate peer-implied fair values of stocks from their accounting items and the market value of other stocks with different machine learning techniques (both linear regression and non-linear regression such as decision tree, random forest, support vector regression, Gradient boosted regression trees). The research has also manipulated input of the data by determining the significance level of the variables using VIF, correlation parameters etc. and filter out useless factors for prediction. Assuming no short-selling constraint or trading costs, divergence between the stock’s market value and peer-implied value represents mispricing and convergence trading opportunities with over 10% annual return. Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: LIU Zhetian / MAEC CHENG Hongyu / FINA YANG Wenting / COGBM Course : UROP1100, Spring UROP1100, Spring UROP1100, Spring Our UROP topic is “Investment Analysis with Machine Learning” and our group job is to do technical analysis. We first did some literature review on papers, and then tried to replicate the paper (Han, Zhou & Zhu, 2016). We notice that there remains differences between a variety of stock species, and specifically, Shenzhen GEM performs better than other stock sectors. Furthermore, some evidence implies that the performance of trend factor is not severely affected when prohibiting short-selling, as a suggestion that it could be applied in A share market where short-selling is relatively difficult. This report will include the whole process of research and concrete analysis of the results. 188 School of Business and Management Department of Accounting Investment Analysis with Machine Learning Supervisor: YOU Haifeng / ACCT Student: WANG Jianghan / FINA Course : UROP1100, Spring We replicate Bartram and Grinblatt (2018) using python and make some improvements. In our research, we implement an agnostic fundamental analysis with minimal data snooping. Rather than applying highly stylized models (e.g., DCF valuation) to estimate fair value, we use techniques such as least squares to estimate peer-implied fair values from the market values of replicating portfolios with the same accounting statements as the company being valued. Consequently, we construct a mispricing signal which represents the divergence of a company’s estimated peer-implied fair value from its market value. Fama-MacBeth cross-sectional regressions are used to test if mispricing signal itself can predict future returns. As an improvement, we use LASSO to select the most influencing factors to construct peer-implied fair value. Information Collection from Annual Reports of Public Companies Supervisor: ZHENG Yue / ACCT Student: IP Ching Stephanie / FINA Course: UROP1100, Fall In this project, we attempt to collect product-related information from the annual reports of US high-tech public companies and get familiar with the online resource of filing that the US public companies file with the Securities and Exchange Commission (SEC). We are given an Excel file listing the 500 US companies’ 10-k filings and we have to search the original 10-k file on the SECEdgar Website based on the CIK number and filing date. In the 10-k files, we extract paragraphs or sections describing the companies’ products and services, particularly from Item 1 (Business) into 500 txt format documents. Information Collection from Annual Reports of Public Companies Supervisor: ZHENG Yue / ACCT Student: KAN Ho Yin / FINA Course: UROP1100, Fall In this project, collection of information of US companies is required. During the information collection process, browsing the SEC Edgar Website, searching the 10-K annual reports of US high-tech public companies and reading Item 1 (Business) are needed. In a bid to gain a better understanding of the product and services these high-tech public companies offered and their developments of product space over years, sections or paragraphs describing the companies’ products and services are extracted and compiled into txt documents for reviews. At the end of this project, a total of 500 documents of product-related descriptions from Item 1 of 10-k filings have been made. 189 School of Business and Management Department of Accounting Information Collection from Annual Reports of Public Companies Supervisor: ZHENG Yue / ACCT Student: LAU Chun Yin / ACCT Course: UROP1100, Fall In this report, the EDGAR platform, which was the main search platform for the annual reports used for this UROP1100B project, is introduced. Descriptions of common filing types on the platform, such as 10-K, 10-Q and 8-K, are provided. Moreover, this report discusses about the findings in the searching process. The situation of listed pharmaceutical companies is specifically discussed, while Minerva Neurosciences, Inc. is used as an example to illustrate the risks faced by pharmaceutical companies in the research and development stage. Lastly, the report explores the reasons why some companies change their company name and are unwilling to provide details about their products and services. 190 School of Business and Management Department of Economics Department of Economics Data Collection and Cleaning for Topics in Labor Economics Supervisor: LI Xuan / ECON Student: YE Muyang / ECOF Course: UROP3100, Fall This is a progress report of the UROP project “Data collection and cleaning for topics in labor economics” supervised by Prof. LI Xuan. Building on previous investigation on multinational wage compression, this study aims to test the robustness of the wage shock effect using individual-level data of multinationals in Brazil. The number of multinationals in this dataset is expanded from 63 to 670. The analysis would confirm the wage pattern influence controlled for individual fixed effect. In this project, I am mainly in charge of cleaning data and filtering firms using Stata based on given codes. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: LI Xinzi / ECOF LI Yaxuan / ECOF YANG Jingdian / MAEC Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall This paper investigates the trade pattern between the firms in Fujian Province, China. Firstly, it explores the trade pattern of the firms of different sizes mainly in three dimensions: the number of firms, the trade volume and the number of transactions. Secondly, it gives some observations of firms’ overtime trade pattern. Finally, it offers ways to find the merging rate of firms recorded in customs data and trade data in Fujian Province. It concludes that all types of firms prefer to trade with small and micro enterprises and the increase of the number of firms is basically due to the “New entry” and “Exit”. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: OU Shiying / ECOF BIAN Ce / MAEC Course: UROP1100, Fall UROP2100, Fall This paper continued the research based on the exploration in the impact of BRI on China’s international trade with the trading partners. It complemented the previous gravity analysis on the bilateral trade among 142 countries in the world during 2010- 2018 respectively. More specifically, the research explored China’s export growth at the intensive margin and extensive margin during 2010-2019 respectively, and concluded that China’s export values mainly grow at the extensive margin, especially in product diversification. Lastly, it analyzed the dynamics of China’s exporting products entry, continuation and exit. Result shows that the income inflow has more potentials to grow and become more stable as the development of BRI policy deepens. 191 School of Business and Management Department of Economics Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: GAO Zimo / ECOF ZHANG Ziyang / MAEC TANG Yunyang / MAEC Course: UROP3100, Fall UROP3100, Fall UROP4100, Fall The United States has implemented a set of tariff policies against China since 2018, leading to the Trade disputes between China and the United States. In the last project in Summer, the authors studied the effect of the US tariff policy on China’s export to the US as well as its diversion effect on China’s export to third countries. In this report, the authors took a closer look at the effect of the UStariff policies and turned to focus on the effect on the trades between other countries other than China and the US. Typically, the authors selected the countries with the largest trade value with China and the US as the target countries. The authors collected the data of import and export value of 2016-2020 between those countries and continuing running regressions to study the effect of the US tariff policy. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: SHENG Ruoheng / MAEC WANG Shuoran / MAEC Course: UROP3100, Fall UROP3100, Fall This paper extends the previous empirical analysis on the impact of a foreign invested firm’s home sourcing dependence and processing trade intensity on the firm’s export performance. Using two datasets, namely Taiwan Invested Firms data (TIF data, hereafter), and transaction data from the Chinese custom (custom data, hereafter), we first performed cleaning on the data from 2000 to 2009 to extend our data samples to a longer period. Meanwhile, we revised our data cleaning methodology for the data samples from 2010 to 2015 to include additional indicators for export performance and revised our regression models accordingly. We finally obtained new evidence that supports our hypothesis on the impacts of home sourcing dependence and processing trade intensity on export performance. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: JIA Ningxin / ECOF WANG Yixuan / MAEC Course: UROP1100, Spring UROP1100, Spring Globalization of companies and industries is such a complex topic that requires great amount of data as evidence to support the research and conclusions. Mainland firms invested by Taiwan parent companies is a kind of cogent evidence among different types of firms. The primary task is to obtain a list of firms which satisfies the requirement of data mining and analysis. 192 School of Business and Management Department of Economics Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: YAU Chin Fung / MAEC Course: UROP2100, Spring I work on the 2018-2019 US-China trade war and the related tech war. In this project, I completed three tasks. The first task is to identify the Harmonized System (HS) codes that are related to tech war and 5G. The second task is to find out the percentage of products which are subjected to an increase in tariff rate. This report will show my code or algorithm and my findings of top targeted products. I will show what HS chapters have a higher percentage. The third task is to convert the list of products which China increased tariff on US-imported products (due to retaliation) to Stata files. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: BIAN Ce / MAEC OU Shiying / ECOF Course: UROP3100, Spring UROP2100, Spring This paper continued the research investigating the change in trade patterns of China and BRI member countries under Belt and Road Initiative (BRI). We polished our gravity model analysis of the bilateral trade with all-country sample, and added a difference-in-differences analysis to test the robustness of our findings. We also improved our analysis on the bilateral trade flow between China and other trading partners by decomposing the total import and export into intensive margin and extensive margin respectively, due to the fact that aggregate level trade flow may hide some detailed changes in trading costs under product level. The results are consistent and expected as our previous findings, showing that BRI has promoted bilateral trade among BRI member countries and China to some extent. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: WANG Shuoran / MAEC Course: UROP4100, Spring With Taiwan-invested firm dataset (TIF dataset, hereafter) and Chinese custom transaction data (Custom dataset, hereafter), we previously discovered the negative effect of home sourcing dependence and the mitigating effect of processing trade on a multinational firm’s export performance. To test the relationship more accurately, we added new Taiwan-invested firms in our data sample and kept new information about supplier base. We also included information about firm size and financial position from Chinese Industrial Enterprise dataset (CIE data, hereafter). After adding new explanatory and control variables in our regression models, we discovered that a larger supplier base will improve a multinational firm’s export performance, while the effects of home sourcing dependence and processing trade were not robust. Keywords: China, Taiwan, trade pattern, regression analysis 193 School of Business and Management Department of Economics Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: GUO Liangyu / ECOF Course: UROP1000, Summer Since its launch in 2013, China’s Belt and Road Initiative (BRI) has enabled the Chinese government to actively invest in nearly 170 countries globally with a focus on large-scale infrastructure construction projects and energy goods. This progress report presents how the Belt and Road Initiative affects China’s different types of trade and OFDI patterns at aggregate and sector level. We found that at aggregate level, countries receiving higher amount of OFDI from China also have high level of trade flow with China. Also, being a initially targeted BRI country will increase the possibility of receiving higher OFDI given the same amount of trade value with China. For energy sector, there also exist positive association between trade flow value and China’s OFDI, and being a BRI country will increase its OFDI given the same amount of China’s import to that country. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: YU Yue / MAEC Course: UROP1000, Summer Production quantity and product type of a firm are significant data in analyzing the firm’s operation. In this report we use the data from the National Bureau of Statistics of China firm-product level production survey that contains physical output quantity information. This report summarizes our group’s findings of UROP 1100 in Summer 2021. Given the productivity data which contains product type, and output quantity, our group aims to find the pattern between firm survival and product scope. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: CHAN Cheuk Yin Rudolph / ECON Course: UROP1100, Summer In this paper, we describe the data cleansing process of the Chinese firm production dataset with the product code classification dataset to identify the hidden products, and construct the product scope, product rank and core product at the level of the most disaggregated products. We find that the single-product firms are the overwhelming majority in the full sample and their proportions have climbed consistently over time. The new entrants and exiting firms show similar pattern until 2006, and the total figures in 2007 and 2009 are significantly smaller than those from previous years. Besides, among the surviving firms, the proportion of single-product firms switching from multiproduct firms are consistently greater than that of the other way around until 2006, and the potential data quality issue in 2007 flip the picture around temporarily. 194 School of Business and Management Department of Economics Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: DUAN Yuting / MAEC WANG Yixuan / MAEC Course: UROP1100, Summer UROP2100, Summer This paper aims to study the impact of the trade war between China and US, especially on global value chain of 5G-related products. We mainly observe and compare the change of import value and applied tariff of China and US. We also take attempt to find their correlation by regression. In an overall view, China's import relies on US much less and reduces relatively quicker after trade war. Meanwhile, there is a negative relation between tariff and trade value. As for 5G-related products, most of them are moderately dependent on each other while China relies more on US. As a result, trade war is reducing both counties' dependence on each other, which in general bring more challenges to US but has opposite effect among 5G-related products. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: LI Yuhe / RMBI LUK Wing Yiu / ECOF Course: UROP1100, Summer UROP1100, Summer This paper focuses on the significance of FDI in raising the productivity of Chinese firms. We deployed a firm-level dataset to deduce the relationship between labour productivity and firms’ ownership. The results show that the productivity of foreign-invested firms is higher than their domestic counterparts. Our results are also robust if we use different definitions of foreigninvested firms. Besides, we summarized the destination and industry distribution, and labour productivity of OFDI firms. We found that the most OFDI intensive destinations are developing countries when measured using capital investment but are developed countries when measured using number of investing firms. OFDI intensive industries are highly concentrated in electronics, transportation and metals sectors. Also, firms with OFDI have significantly higher labour productivity than the ones that do not, especially domestic firms. Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: LIAO Caixing / ECOF SU Xuanjing / ECOF Course: UROP1100, Summer UROP1100, Summer This report documents aggregate level and product level facts regarding the trade war. The first part introduces background information. The second part studies the mutual trade importance of the US and China, as well as their trade diversion destinations. The third part quantifies the trade diversification level. The fourth part examines the US-China trade balance, thus sheds light on the bilateral comparative advantage of product groups classified based on stages of processing. The last part concludes the research and proposes future research directions. More disaggregate level analysis is required to identify the causality between the trade war and trade diversion effects. 195 School of Business and Management Department of Economics Firms in Globalization: Evidence from China Supervisor: LI Yao / ECON Student: XU Runzhao / MAEC Course: UROP1100, Summer Our preliminary goal is to separate single product firms from multi-production firms using product code library dataset and product quantity datasets (firm-product level) from year 2000 to year 2009. After merging product code library and each product quantity dataset using corporation codes, we append all datasets to get a panel data. We then classify surviving firms and type-transforming firms which are further classified as single product firms changing to multi-product firms and multi-product firms changing to single product firms using 1-year and 3-year observation window respectively. Finally, we calculate the product scope of each firm in 2007 and 2009 with hidden layer and without it. The Local Impacts of BRI Projects Supervisor: LIN Yatang / ECON Student: CHEN Yi / ECOF Course: UROP1100, Spring In the context of global climate change, environmental problems have been a widely spread concern all over the world and green economy has also become a globally big topic. Thus with the continuous development of China’s Belt and Road Initiatives (BRI), its impact on the local economic growth as well as environment have become the two biggest concerns. Since economic growth is usually accompanied by huge carbon emissions, how to balance economic development and environmental protection is undoubtedly a vital and crucial issue. By studying the impact of the BRI projects on the local GTFP, we will explore whether BRI can promote local green economy development. The Local Impacts of BRI Projects Supervisor: LIN Yatang / ECON Student: ZHANG Han / ECOF Course: UROP1100, Spring The Belt and Road Initiative (BRI) launched by China in 2013 has been extremely ambitious and brought a tremendous impact on major regions. Yet, concerns over its environmental impact have been raised. Since a large fraction of BRI constitutes energy projects, this paper studies the BRI-driven investment pattern in the energy sector, especially in renewable energy. Using multiple empirical strategies, this paper provides evidence that relative to other sectors, BRI has increased China’s outbound financing in renewable energy to non-BRI countries rather than BRI countries. When further investigating what type of countries would attract more renewable energy funding from China, it is shown that places with a better commitment to renewable energy promotion, as measure by their Nationally Determined Contributions, would draw more investment from China. 196 School of Business and Management Department of Economics Exploratory Research to Inform a Financial Planning Tool for Migrant Domestic Workers in Hong Kong Supervisor: VISARIA Sujata / ECON Student: FU Zixuan / MAEC Course: UROP1100, Summer The project is a pilot survey of a future panel study that follows both migrant domestic workers and their family members. The topics concerned in the exploratory study include immigration goal, financial decision, communication, nutrition and health. At the exploratory stage, a qualitative research was conducted in the form of Focus Group Discussion (FGD), preceding small-scale quantitative research in the form of questionnaire and interview. Knowledge built up from the qualitative research is crucial to the questionnaire design for data collection. However, instead of focusing on the academic insights from the research, this report mainly introduces the author’s experience and takeaway over the course of the project, from both literature review and conversational interview. Exploratory Research to Inform a Financial Planning Tool for Migrant Domestic Workers in Hong Kong Supervisor: VISARIA Sujata / ECON Student: ONGRATANAPHOL Ramon / MAEC Course: UROP1100, Summer Filipina domestic workers in Hong Kong have several difficulties when trying to save as low savings and high borrowing habits lead to poor financial decisions. To explore tools that could assist them to better manage their financials, focus group discussions and quantitative interviews were conducted to understand their current situation. An absence of dedicated savings account in Hong Kong, high remittance along poor communication with family could be a possible reason for explaining these phenomena. This report aims to highlight memorable points that participants have mentioned, contrast these findings with previous research done on Filipina Domestic Worker and attempt to provide possible interventions. Keywords: Filipina domestic workers, financial decisions, financial interventions. Exploratory Research to Inform a Financial Planning Tool for Migrant Domestic Workers in Hong Kong Supervisor: VISARIA Sujata / ECON Student: PURI Sneha / ECON Course: UROP1100, Summer One in three households in Hong Kong have hired a Migrant Domestic Worker (MDW)1. Almost half of them are Filipino, who are referred to as Overseas Filipino Workers (OFWs)2. Over the course of 2 months, we conducted focus group or one-to-one interviews with over 130OFWs, with varying life experiences, family situations, working conditions, and demographics. Through the interviews, we aimed to gain insight into their finances, health, family back in the Philippines, and working conditions in Hong Kong. Personally, I was in-charge of recruiting the OFWs, inputting the survey questions on Qualtrics, and on-the-day logistics. After conducting all the interviews, I was also involved with data cleaning and evaluation of the survey for future improvements. 197 School of Business and Management Department of Economics Kin 'Taxes'? How Filipina Domestic Workers in Hong Kong Manage Family Expectations Supervisor: VISARIA Sujata / ECON Student: TANG Tsz Yeung Alvis / ECOF Course: UROP1100, Summer Filipino domestic workers earn multiples of the salaries in the Philippines, yet more than half of Filipina domestic workers in Hong Kong report no improvement financially after they migrate to Hong Kong. The reasons for the limited developmental impact of Filipina domestic workers are little studied holistically. In Summer 2021, The UROP team conducted focus group discussions and quantitative surveys among Filipina domestic workers in Hong Kong as exploratory work for a future study. In this report, I describe the research motivation, introduce the research questions in mind, outline my role as a research assistant in the project, and reflect on my growth in the research process. Conduct Economic Research for Teaching Purpose Supervisor: YU Yan / ECON Student: HE Kai / SBM Course: UROP1000, Summer Like many other public utilities, the electricity market is a natural monopoly due to its high infrastructural costs. To guarantee affordable electricity prices for consumers, governments have attempted various types of regulation. In Hong Kong, electricity is provided by two vertically integrated monopolies under rate-of-return regulation. In Singapore, electricity was originally provided by a vertically integrated monopoly, like Hong Kong. However, in 1995, the Singapore government began to introduce competition in the electricity sector. Since then, the government has continued liberalization efforts, in hopes of forming a more efficient and competitive electricity market. This case study will provide a brief investigation into the regulatory approaches of Hong Kong and Singapore in order to compare their efficacy. Conduct Economic Research for Teaching Purpose Supervisor: YU Yan / ECON Student: JAIN Saumya / ECON KAUSHAL Kaustubh / ECON ZHOU Kexin / MAEC Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer The objective of this report is to give an overview of a commodity that we all use in our daily lives -electricity. Interestingly, it is a good that is homogeneous in nature and is very hard and expensive to store, hence, its demand must equal the supply. This report explores theoretical concepts such as price determination through the intersection of demand and supply curves in real life, and how the market structure affects the efficiency of the market. The primary focus of this report will be on the wholesale electricity market in Germany, United Kingdom and United States. 198 School of Business and Management Department of Economics Conduct Economic Research for Teaching Purpose Supervisor: YU Yan / ECON Student: NING Zekai / IS RIYAS Eshayat Fathima / MAEC YIP Kwun Chuen / MAEC Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer This case report follows the liberalised electricity markets in Germany, the United Kingdom and part of the USA, specifically Texas. It follows the question of how effective liberalisation of the electricity sector has been in various markets. In the first three sections, the report focuses on how liberalisation affects different parts of the electricity market in each of the three countries. The first three sections look into the trends in market structure, prices charged to residential and industrial consumers and other aspects, pre and post-liberalisation. In the final section, the report focuses on the various similarities and differences in the liberalisation process in these markets. 199 School of Business and Management Department of Finance Department of Finance Trade Shows and Supply Chain Formation Supervisor: BISETTI Emilio / FINA Student: GUO Liangyu / ECOF LAI Cheuk Lam / ECOF LI Junmei / RMBI LOO Chung Wa / FINA WU Kunxiong / QFIN XU Yibei / FINA Course: UROP1100, Spring UROP1100, Spring UROP1100, Spring UROP1100, Spring UROP1100, Spring UROP1100, Spring Money laundering often hinges upon serious crimes to hide illegal cash flows. In the 1980s, some regional organizations, such as the European Union and the Organization of American States, came up with a global set of measures to avoid this issue among Western countries. As one of the fastest-developing regions, Asia has also increased its anti-money laundering (AML) measures over the past few years. This report aims at examining the effectiveness and stringency of AML laws among 19 major Asian Development Bank members. It relies on 40 Recommendations from mutual evaluation reports of The Financial Action Task Force (FATF) to develop a rating system. Concerning stringency, the report also builds regression models to assess the impact of AML legislation on international trade and economic growth in both developed and developing regions. It indicates that the region's economic performance has improved since the AML law was changed. Trade Shows and Supply Chain Formation Supervisor: BISETTI Emilio / FINA Student: SHYONG Tien Chi / QFIN WU Chen / FINA YAN Ely / FINA ZHANG Zirui / ACCT Course: UROP1100, Summer UROP1100, Summer UROP1100, Summer UROP1100, Summer FATF (Financial Action Task Force) is the most influential anti-money laundering and anti-terrorist financing organization in the world. The organization regularly evaluates the effectiveness of anti-money laundering efforts in various countries. Before the fourth round of FATF evaluation in 2019, many countries in Asia had revised their anti-money laundering laws. The purpose of this report is to explore whether the legal amendments in the 10 Asian countries were well-planned or acted unexpectedly. After reading the relevant news and government reports, we believe that only the legal revisions in Taiwan and India were unexpected, and the legal revisions of the other countries were all well-planned. The report discusses the actual situation of each country (region) on a country-by-country basis. 200 School of Business and Management Department of Finance Climate Policies and Cost of Financing Supervisor: GOYAL Vidhan Krishan / FINA Student: BALDWA Anushka / BIBU Course: UROP1100, Spring The project examines how green bond issuances benefit the issuers. I obtained data on green bonds issued in Hong Kong from the Blomberg Terminal. I then analysed the data to classify issuers into various types such as Corporate Banks and Commuting services and then evaluated the time-series variation in the number of green bonds issued and their maturity dates. The report also summarizes coupon rates and other characteristics of green bond issuers. Finally, I estimated yield spreads on green bonds to test for the presence of ‘greenium’ or green bond premium. After analyzing secondary data on green bonds and their issuers, the next step is to read through green bond prospectuses to understand better the provisions included in green bond issues. Climate Policies and Cost of Financing Supervisor: GOYAL Vidhan Krishan / FINA Student: CHENG Yanxiang / FINA Course: UROP1100, Summer This research aims at exploring the effect of social attention on green bond pricing in China. To understand the underlying mechanism, I dissect social awareness into public and analyst attention to see how they affect bond pricing differently. I then use ordinary least squared regression to test for different attention sources’ significant impacts on green bonds’ spread. The findings indicate that both social and analyst attention plays a vital role in affecting bond pricing in China. Climate Policies and Cost of Financing Supervisor: GOYAL Vidhan Krishan / FINA Student: JIA Yuanchen / ECOF Course: UROP1100, Summer This report examines the investment and green performance of Chinese green mutual funds vis-à-vis their conventional peers. I compare fund return and portfolio average ESG rating and conclude that Chinese green funds exhibit higher returns and have better ESG scores than their conventional peers, suggesting that they are both doing good and doing green. 201 School of Business and Management Department of Finance Climate Policies and Cost of Financing Supervisor: GOYAL Vidhan Krishan / FINA Student: RACHMAT Farah Darlene / FINA Course: UROP1100, Summer Previous research papers have shown that a single firm can have varying ESG ratings from different ESG rating agencies. The resulting information discrepancies have caused the overvaluation of the firms’ stocks, resulting in lower portfolio returns. Besides suboptimal returns, these ESG Indices are also utilized by various fund managers in managing trillions of assets under management. In this report, I analyse the types of Green ETFs that maximize investors’ returns. Furthermore, I also analyse the investment strategy of these Green ETFs alongside its constituents. With these findings, I aim to develop an investing strategy that minimizes the effects of information asymmetry due to disagreements from various ESG ratings. Climate Policies and Cost of Financing Supervisor: GOYAL Vidhan Krishan / FINA Student: XU Tunan / GBM Course: UROP1100, Summer In recent years, ESG disclosure has become the focus in the financial world. Many states legislate to make ESG disclosure mandatory, while it is still a debatable question for governments worldwide to discuss. This paper will examine whether ESG disclosures reduce credit spreads. Credit spread reflects the issuer risk in credit markets, partly reflecting the information asymmetries between firms and bond investors. Higher ESG disclosures can alleviate the information asymmetry and allow investors to understand better legal and regulatory risks arising from ESG. A negative correlation between ESG disclosure score and credit spread is expected for companies when controlling for pollution level, leverage, firm size, and other factors. Climate Policies and Cost of Financing Supervisor: GOYAL Vidhan Krishan / FINA Student: XU Yan / QFIN Course: UROP1100, Summer Among a variety of emerging green financial instruments over the years, green ABS is an innovative vehicle driving the growth of the green bond market. In this paper, I analyze the green ABS market in Mainland China. The report defines and classifies green ABS products. It further provides certification and disclosure requirements. In addition, the report highlights the comparative advantages of green ABS issuances. It then observes trends of green ABS issuances in the Mainland and identifies policies and regulations boosting the green ABS business. Furthermore, it includes a case analysis of Industrial Bank Co., Ltd, focusing on the first labelled green credit ABS issued in Mainland. 202 School of Business and Management Department of Finance Corporate Risk Management Practices: Global Survey of BRICS-Country Firms Supervisor: MACKAY Peter / FINA Student: WANG Jianghan / FINA WANG Matthew Sebastian / ACCT Course: UROP1100, Fall UROP1100, Fall Corporations take different measures to manage their financial risk based on the stakeholders’ needs. The goal of this project is to analyze the impact of these strategies. Currently, the survey stage has already been executed, with results already having been received; the survey results will be analyzed with respect to the daily adjusted stock price to evaluate the impact of the risk management decisions made. The stock prices will be measured against a benchmark of the national stock index, bond prices, and the price of precious metals to isolate the impact of the firms’ financial decisions. At this stage of the project, we have completed data collection and have initiated data pre-processing. Risk-Management Duel: Financial vs Operational Risk Management Supervisor: MACKAY Peter / FINA Student: DENG Wenxuan / RMBI LUO Beier / RMBI TANG Zexi / MATH-SFM ZHANG Qianjiao / RMBI Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall UROP1100, Fall In normal company operation, it is common for companies to observe their financial performance to make diligent development decisions. Companies should always look for expansion opportunities of down-scaling options based on its financial profitability. In this project, financial and operational management data of gold mining companies over the past 20- 30 years will be studied to identify the risk management relationship between these two areas. Now, this project is still in its exploratory stage and now only data relevant to the companies’ operation for example costs and production is collected and stored in a database. 203 School of Business and Management Department of Finance Risk-Management Duel: Financial vs Operational Risk Management Supervisor: MACKAY Peter / FINA Student: AMADOR JAIME Jose Luis / CPGBM HE Jinghui / ECOF ZHAI Menghe / QFIN WANG Matthew Sebastian / ACCT Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer UROP1100, Summer The operations of companies require a mixture of operational risk management and financial risk management. In general, companies look for expansion opportunities or downscaling options based on their financial profitability. This project seeks to analyze the financial and operational management data of mining companies and investigate the interactions of operational risk management and financial risk management. This data includes data over the past 20 years on the characteristics of mines such as production costs, production levels, proven and probable reserves. This data will be analyzed to test whether a company’s portfolio of mines (assets) determines how it uses tools to manage financial risk. Currently, this project is at its early exploratory stages and the scope of this report will describe how data on the mines reserves and production levels were collected. Analyzing CEO Compensation Contracts Supervisor: ZALDOKAS Alminas / FINA Student: YAU Hoi Ching Flora / FINA Course: UROP1100, Fall The project focuses on discussing how the design of CEO compensation incentivizes CEOs to participate in market collusion. In particular, we look into the closure of four regional antitrust offices of the United States Department of Justice Antitrust Division in 2013 and its consequences that decreased the antitrust enforcement. In other words, shareholders and management are motivated to collude with rival firms. We found that the executives' pay is more positively correlated with close competitors. My assigned tasks are to identify the CEO compensation metrics in terms of the company's profitability and also research the case of the closure of four regional offices of DoJ from social media and antitrust community aspects. Products, Supply Chains, and Finance Supervisor: ZALDOKAS Alminas / FINA Student: LI Mengyi / QFIN TU Bingying / DSCT WINATA Timothy / FINA WONG Chun Ho / QFIN Course: UROP3100, Fall UROP1100, Fall UROP1100, Fall UROP1100, Fall The ultimate goal of this research is to find out relationships between New Product Announcement (NPA) and the abnormal return for given stocks listed on NYSE and NASDAQ. In addition to this, we will continue discussing the relationships of the NPA and their effects on the economy. The first stage of this research is to build up the database and utilize big data technologies to clean the database and identify keywords related to the company and their new products. After cleaning up the database, we would include more information to eliminate market-in-build expectations about the product. Finally, we would analyze the stock return based on the database as well as regression models. 204 School of Business and Management Department of Finance Products, Supply Chains, and Finance Supervisor: ZALDOKAS Alminas / FINA Student: LI Mengyi / QFIN Course: UROP4100, Spring The goal of the research is to quantify the relationship between the first-time released New Product Announcement (NPA) and the abnormal return of the underlying stock within a two-day event window. We focused on the implementation issues based on the ideas generated in the previous semester, and we mainly accomplished the following tasks: automatically converted the raw data in .rtf form into .xlsx form, automatically converted the word document in .docx form into .txt form, and retrieved information needed for further processing, and conducted Term Frequency – Inverted Document Frequency (TF-IDF) test to find out similar articles. Products, Supply Chains, and Finance Supervisor: ZALDOKAS Alminas / FINA Student: HUI Siu Lun / ECOF LING Xiuqi / ECOF XIAO Yunyi / QFIN YUAN Xuzhi / FINA WONG Chun Ho / QFIN Course: UROP1100, Spring UROP1100, Spring UROP1100, Spring UROP1100, Spring UROP2100, Spring The ultimate goal of this research is to find out relationships between New Product Announcement (NPA) and the abnormal return for given stocks listed on NYSE and NASDAQ. In addition to this, we will continue discussing the relationships of the NPA and their effects on the economy. One important stage of this research is to build up the database and utilize big data technologies to clean it and identify keywords related to the company and their new products. We exchanged different ideas to overcome the difficulties in this stage, and have been able to set up a framework for further experiments. Next, we have designed codes for keyword identification and extraction is implemented, and we will try refining the code. Products, Supply Chains, and Finance Supervisor: ZALDOKAS Alminas / FINA Student: LUI Hiu Lum / ACCT SHI Qingqing / ECOF ZHUANG Nanpeng / QFIN YUAN Xuzhi / FINA Course: UROP1100, Summer UROP1100, Summer UROP1100, Summer UROP2100, Summer This research will continue the process from last semester to find out the relationships between New Product Announcement (NPA) and the stock return. We will exam whether the second or third news describing the same product can also affect the stock price. We will further analyze if the stock return is correlated with the similarity of 2 news articles or the number of days between 2 articles. In this program, we start from code refinement and data cleaning to abstract all data we need from Bloomberg Terminal. And end up with regression model analysis to make our conclusion. 205 School of Business and Management Department of Finance Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: LUANG Sin Kei / FINA Course: UROP1100, Fall Whether Hong Kong initial public offering firms (IPOs) outperform their peers remains a puzzle. In general, the 605 Hong Kong IPOs listed during 2014-2019 slightly outperform their matching firms in the 1-3 year holding period from the closing price on their IPO date to their 1 year, 2 year, and 3 year anniversaries. There is almost no bias with ±0.01𝜎 towards IPO return for the one year holding period. This bias increases over time and there is a large increase in (mean) return between holding for 1-year and 2-year, then a moderate increase between 2-year and 3-year. The null hypothesis of the paper is that firms with higher EPS have greater long term holding period return (2-3 years). Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: MU Di / ECOF Course: UROP1100, Fall As used in many asset pricing models, firm size is considered as a determinant of expected stock return. However, there isn’t much preliminary research done using the data of the Hong Kong Stock Exchange. We collected data of 247 firms that had their initial public offering issued from 2014-2016. Then we computed their return in 1-month, 3-month, 3-year respectively and took the difference between those and the corresponding market/industry return. We found evidence of the size effect of larger companies are associated with smaller returns using both benchmark portfolios. The results are consistent in all three time periods. Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: SHEN Yue / IS Course: UROP1100, Fall The research focuses on the Hong Kong initial public offerings listed between 2016 to 2019. Through the statics analysis of the excess return of the Hong Kong IPO, it shows that the underpricing phenomenon is common in the short term. Through the utilization of the ordinary least square model, the EPS is positively correlated with the excess return of the Hong Kong IPO. 206 School of Business and Management Department of Finance Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: LAM Ching Long / ECOF NGAI Ho Fai Calvin / QFIN Course: UROP1100, Spring UROP1100, Spring This research project aims to find out a model to predict the offer price of IPOs in HKEx, given the publicly available information right before the listing. Using financial data consolidated from IPO prospectuses and DataStream, two distinct models for price prediction were built using Microsoft Excel and Python. Factors regarding company size, time and market sentiment indeed have a significant effect on the IPO offer price. A counterintuitive finding is that the price-to-book ratio of the industry at the time of issue has a negative relation with the IPO offer price. The methodology of the research and the results of the models will be elaborated in detail in the report. More other important findings and implications of the results would also be discussed. Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: LIU Muyuan / QFIN Course: UROP1100, Spring Before listing a company on Hong Kong Stock Exchange, companies need to hold a roadshow to collect information from the institutional investors to help with the offer price determination. However, there hasn’t been a model which reveals the relationship between IPO offer price and the underlying factors. We gathered the data from 2772 companies that got listed on Hong Kong Stock Exchange, whose date of listing range from 2002 to 2020. We finally find that the firm’s offer price is associated with the company’s debt ratio, return on equity, return on asset, total equity, industry average equity value, and the industry-average market-to-book ratio. The model shows a good prediction accuracy in the data. Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: SUN Beini / RMBI Course: UROP1100, Spring The goal of this research is to find out a model that can describe the offer price given information from the prospectus. We have built two kinds of prediction models, linear regression models and nonlinear (decision tree) models to observe which factor is likely to have an impact on the IPO offer price. Findings from all the linear models we have constructed are that the financial fundamental ratios (ROA ROE and Equity Multiplier) in the latest year tend to have largest coefficients and smallest p-value, suggesting that the most recent financial status play a vital role in IPO pricing. For nonlinear models, the most informative feature is the average ROA-ratio over the past three years. 207 School of Business and Management Department of Finance Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: WANG Yiran / QFIN ZHAO Xuehan / QFIN Course: UROP1100, Summer UROP1100, Summer The research aims to find out a model that describes the IPO price in the Growth Enterprise Market (GEM) of the Stock Exchange of Hong Kong as a function of the information variables known at the time of the offer. We investigated the factors that help investors determine the IPO price at the time of the offer and analyzed how strong each of the relationships is. We used financial ratios, indices, market size indicators, and industry classification variables of each company and applied these variables into three model selection methods. The result shows that the stepwise selection model gives the best performance. And based on this model, we provided a reasonable explanation for each of the factors involved. Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: WONG Shing Bun / QFIN Course: UROP1100, Spring The purpose of this study is to discover and quantify the offer price-setting behaviors of Investment banks engaged in the Hong Kong IPO market in terms of fundamental, technical, market, and psychological aspects. From the linear regression model built during the study, I discovered that financial data 2 years before the IPO especially data involving asset usage efficiency, as well as the timing of the IPO plays a significant role in determining its offer price. It may be attributed to the mechanism of fair value changes of convertible preferred shares and possible net income manipulation concerns from the underwriters and investors. Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: SHEN Yue / IS Course: UROP2100, Spring The research focuses on the Hong Kong IPO listed between 2002 to 2020. Through the statis analysis of the market to book ratio of the Hong Kong IPO, it shows that the companies with higher market to book ratio has common attributes. They have relatively lower fund raised, higher gross spread, higher placing share proportion and higher net income of the year of listing. 208 School of Business and Management Department of Finance Hong Kong IPO Market Supervisor: ZHANG Chu / FINA Student: LI Zhuoyun / ECOF Course: UROP1100, Summer An IPO is a significant stage in the growth of many businesses, as it provides them with access to the public capital market and increases their credibility and exposure. The determination of the offering price is one of the key stages in the IPO. In addition to the demand for a company's shares, there are several other factors that determine an IPO valuation, including industry comparable, growth prospects, and the narrative of a company. By studying the relevant prospectus and running the models, it shows the recent Equity to Asset ratio is the most important feature to determine the companies P/B ratio in IPO. 209 School of Business and Management Department of Information Systems, Business Statistics and Operations Management Department of Information Systems, Business Statistics and Operations Management Algorithmic Trading Strategies Supervisor: CHEN Yanzhen / ISOM Student: PANGESTU Dani Surya / ECOF WONG Hon Yu / OM ZHOU Zhuorui / COGBM ZHU Xinyu / QFIN Course: UROP1100, Fall UROP1100, Fall UROP1100, Fall UROP3100, Fall Our group accessed and preprocessed the real-time market data during the meeting of the Federal Open Market Committee Press Conference from 2011 to 2017. Then these data are regressed between the market indexes and the speaker behavioural indicator. Behavioural indicators include factors such as blink rate while the market index includes e-mini futures rate, VIX and so on. This serves as an attempt to explore whether the behavioural indicator of the speaker can send market movement signals or not, and is there any chance to construct trading strategies if any pattern of market reaction according to the speaker behaviors is found based on the regression result. Financial Analyses of Tech Companies Supervisor: KOH Tat Koon / ISOM Student: TANG Zexi / MATH-SFM Course: UROP1100, Spring This project aims to examine tech companies' (e.g., FAANG BAT, Salesforce, Oracle, etc.) strategy and financial performance. By collecting weekly stock prices from Bloomberg and analysing stock return with special regression tactics, this project discovers the relationship between mainland tech companies and tries to predict their stock prices in near future. Now, this project is still in its exploratory stage and now only data relevant to mainland tech companies (Alibaba, Meituan, Pinduoduo, Vipshop and Maoyan) is collected and stored in a database. 210 School of Business and Management Department of Information Systems, Business Statistics and Operations Management Applicability of Deep Learning Solutions to Business Problems Supervisor: KWOK James Sai Ho / ISOM Student: YEUNG Man Yin Michael / MAEC Course: UROP1000, Summer This paper investigates the selection of specifications of deep neural networks (DNNs) for binary classification in business problems, including input dataset dimensionality and model structure. In this study, specifications of DNN models designed for different business applications are used as the empirical data for analyses. These DNN models are tuned to have test accuracies optimized as possible by varying specifications. The results showed that the number of features, number of training samples and number of neurons are important factors having positive relationships with test accuracies. The relationships are modelled by an asymptotic regression model. The results provide a guideline for building DNN models for business binary classification problems and an estimator is proposed for the anticipation of the test accuracies of such models. Gender-Biased Perception Towards Entrepreneurship Supervisor: KWON Ohchan / ISOM Student: ARISTIA Joan / QFIN WONG Tsz Kwan / IS Course: UROP1000, Summer UROP1100, Summer In comparison to male entrepreneurs, the number of female entrepreneurs is far below. This indicates that there is a gender gap in entrepreneurship. In this research, we examine whether the gender gap is influenced by a gender-biased perception toward entrepreneurship. By collecting data from the American business reality television series, Shark Tank, we investigate if female entrepreneurs were particularly disadvantaged in getting investments from the investors. After the data collection process, the data is analyzed with R-programming language. The result of the analysis suggests that female entrepreneurs are at a disadvantage compared to male entrepreneurs. So the case of gender-biased perception toward entrepreneurship is apparent. This conclusion is derived from the value of the company and the number of bids received by female entrepreneurs which is less than male entrepreneurs. Gender-Biased Perception Towards Entrepreneurship Supervisor: KWON Ohchan / ISOM Student: QUACH Tuan Kiet / SBM TAN Chen-yi / SBM Course: UROP1000, Summer UROP1000, Summer Compared to male entrepreneurs, female entrepreneurs need to overcome biased evaluation. In this research, we examine if female entrepreneurs frame their entrepreneurial motivation differently than male entrepreneurs, and whether such differences in framing strategy contributes to reducing the bias. We examine our research question in the context of Shark Tank, a popular reality TV show in the United States. Our initial results suggest that female entrepreneurs emphasize their user experience more frequently than male entrepreneurs, even within the same industry. To understand the impact, we plan to examine the implications for fundraising by employing micro-level data and conduct text analysis to see how framing as user-entrepreneur shapes the follow-up questions from investors. 211 School of Business and Management Department of Information Systems, Business Statistics and Operations Management Identifying Entrepreneurs' Storytelling Strategies Using Twitter Data Supervisor: KWON Ohchan / ISOM Student: FANG Yuhan / ECON Course: UROP1100, Spring The boosting of IT infrastructure and digital platforms has dramatically changed the economic world as well as our everyday life, and much research has been done to detect the impacts of IT infrastructure on regional wages and the impacts of digital platforms like Craigslist, Uber, Lyft, Airbnb, Yelp and so on. Although there is much research on how the Internet affects economic performance in a broad sense, we rarely focus on their impacts on some specific companies from a concrete point of view. To establish this, we present social media data and identify entrepreneurs’ storytelling strategies from the data in order to further explore the association between Internet involvement and business development. Who Becomes Venture Capitalists? Supervisor: KWON Ohchan / ISOM Student: FUNG Kei / ACCT LAU Wing Lung / FINA LUANG Sin Kei / FINA TSE Kai Hei / FINA Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer UROP1000, Summer “Who Becomes Venture Capitalists?” is an interesting topic because the venture capital industry has the stereotype about being dominated by the male. We are curious if this is really the truth nowadays and therefore decide to do a series of analysis on the industry. On Identifying Antecedents and Consequences of Corporate Lobbying Supervisor: LEE Dongwon / ISOM Student: LEE Sehun / OM Course: UROP1000, Summer Since 1995, corporate lobbying activity has been legal in the United States, bringing transparency to the corporate’s legislative approaches. Looking at the data of corporate lobbying in the U.S., Big tech firms such as Google, Facebook, Apple have increased their weight on corporate lobbying as well as CPT lobbying. The fact raised a question; why do those tech firms do CPT lobbying even though they are growing exponentially and what is the effectiveness? In this paper, I closely looked into the CPT lobbying amount of big tech firms and the acts that tech firms directly supported through lobbying, specified on the lobbying disclosure form. 212 School of Business and Management Department of Information Systems, Business Statistics and Operations Management The Development of Innovation and Technology Industry in Hong Kong Supervisor: XU Yan / ISOM Student: CHAN On Yu / IS Course: UROP1100, Fall This report is written for the submission as progress report of UROP 1100 in The Hong Kong University of Science and Technology. Recent publications have highlighted the urgent need of developing Hong Kong’s Technology and Innovation (I&T) Industry. Governments play a crucial role in improving economies' innovation capacity. Effective government policies have proven to be the foundation for fostering development of prosperous I&T ecosystems. Yet, when compared to many other developed economies, Hong Kong is a relative late starter in promoting the I&T industry. The report tries to shed light on Hong Kong’s I&T industry performance and proposes policies to foster industry growth. The information and data presented in this report has been collected from primary and secondary sources. The Development of Innovation and Technology Industry in Hong Kong Supervisor: XU Yan / ISOM Student: TSO Chin Lap Justin / BIBU Course: UROP1100, Fall In the past decade, a series of policies, initiatives, funding, and bureaus were set up to foster the development of the Innovation and Technology industry in Hong Kong are formulated and implemented. With the proposition of the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) in 2017, Hong Kong is looking forward to capitalizing on the opportunities brought by such a megalopolis. Based upon policy address and legislative council keynotes, policies and industrial initiatives on Hong Kong’s technology and innovation within the context of GBA could be divided into four main segments: (1) Foster business and innovation collaboration, (2) Construct and integrate R&D and incubation infrastructure, (3) Nurture youth talent and entrepreneurs to anchor regional development The Development of Innovation and Technology Industry in Hong Kong Supervisor: XU Yan / ISOM Student: WONG Hor Kei / IS Course: UROP1100, Fall Hong Kong’s innovation and technology development has been lagging behind its counterparts, despite being an international hub with a developed economy. With the wave of industrial revolution 4.0 (or “IR4.0”), featuring artificial intelligences, automation and further integration and connection between machines, there may be a golden opportunity for Hong Kong to position itself to be international innovation hub and hence improve social mobility, by joining hands with the Mainland government, industries, academia and talents. The paper is a progress report exploring the impacts brought by IR4.0, followed by a SWOT analysis discussing how Hong Kong can leverage IR4.0 with the focus of local teenage mobility. 213 School of Business and Management Department of Information Systems, Business Statistics and Operations Management Deep Learning NLP in finance Supervisor: YANG Yi / ISOM Student: SUN Ruijie / FINA TAM Shing Hang Boris / COGBM Course: UROP1100, Spring UROP1100, Spring The project is about analyzing hedge words from companies’ conference calls. A hedge word in Finance means a speaker trying to reduce the risks of providing false information in their speech by including uncertainties in their sentences. For instance, an example about hedge words can be “may”, “probably”, etc. Having different hedging levels in conference calls could mean the uncertainty level of that company, hence impacting its values and prospects (negatively or positively). In our project, the aim was not correlating between hedging levels with the corporate performance; but instead to utilize the Natural Language Processing techniques to analyze the hedging levels of a particular conference call based on its speakers’ transcript. Deep Learning NLP in finance Supervisor: YANG Yi / ISOM Student: JIA Tongtong / QFIN Course: UROP1100, Summer BERT, which stands for Bidirectional Encoder Representations from Transformers, has made significant progress in helping computers completing many kinds of NLP tasks by using surrounding text from both directions to establish context. Based on the original version of BERT, there have been many domain-specific BERT models fine-tuned with a specific corpus, which are better suited and perform better when dealing with tasks in specific domains, such as FinBERT (McCormick, 2021). However, we lack some fine-tuned FinBERT models which focus on Chinese finance corpora. To address this issue, we present Chinese FinBERT which is a financial domain-specific language model fine-tuned on large-scale Chinese financial corpora. The results show that our model outperforms many current models with high effectiveness and robustness. 214 School of Business and Management Department of Marketing Department of Marketing Luxury Marketing in China’s Live Streaming Market Supervisor: HELSEN Kristiaan / MARK Student: YU Huixi / ECON Course: UROP1100, Fall As COVID 19 pandemic was hitting the global economy, the luxury industry has suffered great loss. After observing the thriving live-streaming market in China, some luxury brands found it a precious chance for launching their promotion on online platforms. However, the main characteristics of live-streaming seems to confront the positioning of luxury brands. Although the measure could be considered as a temporary strategy to recover profits, we are seeing future possibilities for luxury brands to take entering this new form of business as their long-term strategy. Therefore, the present work aims to analyze reasons which might affect people watching luxury brands’ live-streaming. To do so, a survey was conducted and the results are still in the process of analysis. Luxury Marketing in China’s Live Streaming Market Supervisor: HELSEN Kristiaan / MARK Student: YU Huixi / ECON Course: UROP2100, Spring As the luxury industry has suffered great loss during the COVID-19 pandemic, some of the brands have found it an opportunity to fight back through the Chinese Ecommerce live-streaming. In view of the positioning contrast between luxury brands and live-streaming, it is important to understand the factors that affect people’s willingness to watch luxury live-streaming before developing long-term strategies. To do so, a questionnaire was conducted and the result revealed that the potential audience may have a broad purchasing behavior distribution, and the will was influenced by the information acquiring behaviors and live-streaming utility expectations as well. We suggest that the diverse characteristics and needs of the audience should be the key focus for luxury brands while designing live-streaming. 215 School of Business and Management Department of Management Department of Management Multiculturalism, Social Judgments and Decisions Supervisor: CHAO Melody Man Chi / MGMT Student: GUNADI Vanessa / ACCT Course: UROP1100, Fall The implications of psychological intervention to burnout, mental health syndrome, psychosomatic symptoms, implicit cultural perception of burnout, and conflict management strategies will be discussed in this report. Researchers hypothesized that psychological intervention such as expressive writing would reduce burnout, mental health syndromes and psychosomatic symptoms. Furthermore, after the psychological intervention, researchers predict that people might change their implicit cultural perception of burnout. Researchers will also conduct exploratory research on how psychological intervention will change the conflict management strategies that people will opt for under burnout. Under a burnout situation, people will have lower mental resources, so they will use obliging or avoiding conflict management strategies. Multiculturalism, Social Judgments and Decisions Supervisor: CHAO Melody Man Chi / MGMT Student: KIM Hayeon / CEGBM Course: UROP3100, Summer In modern society, deterioration of mental health is becoming a serious problem because related symptoms including depression, anxiety, stress, and burnout bring upon worsened health conditions and greatly hinder daily activities. However, due to social stigma and high costs, many avoid going to a clinic or even just going for counseling. Therefore, there is ongoing research on finding the most practical, affordable yet effective way to maintain good mental health. Therapeutic writing intervention in particular has grabbed an attention of the researchers as it is an easy intervention where writers can experience catharsis through freely writing about what has been bothering them. Due to a simple instruction and nothing more than writing honestly about what they are feeling, the writing intervention has a potential to serve as an easily accessible, time-efficient and cost-effective therapeutic method. This UROP project investigates the effectiveness of such writing intervention and explore whether it is affected by individual characteristics. 216 School of Business and Management Department of Management Social Perceptions, Justice, and Diversity Supervisor: CHAO Melody Man Chi / MGMT Student: CHEN Jingruo / QSA Course: UROP1100, Spring The Covid-19 has led to large-scale unemployment and changed the working mode significantly, imposing tremendous pressure on employees, which might be correlated with their attitudes towards work. In addition, evidence suggests that the pandemic may have affected women more severely in the workplace with the blurring line between work and life. In this project, I investigate the correlation between wellbeing, work ethics and gender by exploring data retrieved from World Value Survey. I also help design and analyze surveys exploring people's views towards employees who are work-focused or prefer work-life balance under specific scenarios such as behaving unethical pro-organizationally or being interrupted when working home. Social Perceptions, Justice, and Diversity Supervisor: CHAO Melody Man Chi / MGMT Student: PAN Xinyi / ECOF Course: UROP2100, Spring The proliferation of technology has allowed Artificial Intelligence (AI) to take up various roles in organizations. While some are hopeful about the increase in efficiency AI can bring to the workplace, others feel threatened about their increasing presence in the workplace. We aim to examine the impact of such contrasting perceptions of AI in an intergroup context: How does one’s perception of AI affect their prosocial behavior, particularly towards ingroup and outgroup members? Whereas previous studies had inconsistent findings on the impact of AI on intergroup relations, our study (N = 279) found that perceiving AI as a threat (vs. a benefit) leads to more prosocial behavior towards outgroup members. Theoretical and practical implications of the findings are discussed. Social Perceptions, Justice, and Diversity Supervisor: CHAO Melody Man Chi / MGMT Student: JHA Spriha / MGMT Course: UROP1000, Summer Emojis have become an increasingly important part of today’s youth’s lives. Hence, it is both essential and interesting to study how these emojis are perceived by different age groups, genders, and backgrounds. Having enrolled in UROP1000 to find out more about the field of academic research, 16 articles from the existing literature about emojis are reviewed and summarized. The findings of these articles reveal themes on why people use emojis and how they affect their perceptions and behaviors. People mostly use emojis to make up for the lack of the body language element in virtual communications. Emojis are usually perceived as incompetent in formal settings and create a negative bias towards the sender. The research articles also present some gaps, which serve as future directions for further research. 217 School of Business and Management Department of Management Social Perceptions, Justice, and Diversity Supervisor: CHAO Melody Man Chi / MGMT Student: JUNG Jaewon / SBM Course: UROP1100, Summer This report will cover mainly about emotional display and how emotional display would work in a digital context using emojis. The main objective of this report is to explain how emotional display would affect communication patterns especially in negotiation context and how anger would lead to a change of behavior, following with my personal thoughts of future directions. Further on, with emotional display also having an effect on digital context, this report would talk about the overall function of emojis and how emojis could also work as expressing emotions. This report would also talk about various moderating roles of demographic characteristics in terms of social status and gender. Social Perceptions, Justice, and Diversity Supervisor: CHAO Melody Man Chi / MGMT Student: CHEN Jingruo / QSA Course: UROP2100, Summer Fatherhood premium refers to the positive advantages in careers related to men who have children. The main explanations in the literature of fatherhood premium are the specialization perspective and employers' perceptions and discrimination perspective. This literature review aims to bring together information on whether marital status, working hours and the fathers’ race are influential factors of the fatherhood premium and whether the variation is related to or might partly support the two explanations using relevant publications in scholarly journals. 218 Abstracts of UROP Projects 2020-21 School of Humanities and Social Science School of Humanities and Social Science Division of Humanities Division of Humanities Korean War Prisoners Oral History Interview Transcription Supervisor: CHANG David Cheng / HUMA Student: HUI Nga Yin / BIBU Course: UROP1000, Summer The Korean War was initially a war between South Korea and North Korea when it first broke out on June 25, 1950. It soon became an international war when the United Nations forces intervened days later. As North Korea received military support from China, more than one million Chinese soldiers were involved in this war. Professor David Cheng Chang has interviewed more than 90 Korean War veterans, including He Hanchen (何漢臣), a Chinese prisoner of war (POW) who refused repatriation and went to Taiwan. My duty is to transcribe the interviews of He conducted by Professor Chang from video clips. In this report, I would share my research experience in this oral history transcription project. Korean War Prisoners Oral History Interview Transcription Supervisor: CHANG David Cheng / HUMA Student: DONG Bochen / GCS Course: UROP1100, Summer Mr. PAN Dingrong was in his nineties when he was interviewed by Professor Chang and his partners in 2018. He served as a medical officer in the army for many years. He narrated his early life and what he witnessed during the Korean War. In this UROP, I managed to transcribe every word the interviewee said into a Word file. After listening repeatedly to check its accuracy, I summarized and translated the document into this final report, which reveals Mr. Pan's experiences in the Battle of Chosin Reservoir based on his recollections. Korean War Prisoners Oral History Interview Transcription Supervisor: CHANG David Cheng / HUMA Student: XUE Yuyuan / GCS Course: UROP1100, Summer Shortly after Japan’s attack on Pearl Harbor, China and the United States became allies in World War II. China formed the Chinese Expeditionary Force to cooperate with the British and American forces in the China-Burma-India Theatre (CBI). As the number of the American personnel coming to the theatre soared, the number of interpreters needed in the Chinese Expeditionary Force also increased. This research project conducted by Professor Chang David Cheng is precisely about the role these Allied interpreters played in the war, as well as their personal experiences. My task is to transcribe the video interview Professor Chang conducted with an Allied interpreter, Dr. Chen Wen-Chao (陳文昭), in 2012. This report contains details of my work process, collection of relevant literature, and a brief reflection. 220 School of Humanities and Social Science Division of Humanities Building Docent Tours on the Community Heritage of Yimtintsai Village, Saikung Supervisor: CHEUNG Siu Woo / HUMA Student: WONG Sik Yu / GCS TAM Sum Yee / GCS Course: UROP1100, Fall UROP1100, Fall UROP2100, Spring The Hakka village on Yimtintsai Island in Saikung is famous for its UNESCO award-winning heritage projects of the renovated village chapel and the revived salt pans. Yet the history and daily life of the local villagers before they abandoned the island in the late 1990s has yet to be explored for the further development of the island’s heritage program for visitors. This project is intended to be the continuation of the UROP 1000 (Building Docent Tours on the community heritage of Yimtintsai Village, Sai Kung) conducted in Summer 2020, with the aims of producing necessary materials for building the docent tours and docent training workshops in the future. On-site fieldwork and interviews with villagers are conducted in both Summer 2020 and this semester (Fall 2020). We devised the docent narrative script for the tour around the Little Alley, one of the former residential areas of the village. The filming project was not conducted as expected due to the aggravating conditions of the COVID-19 but we managed to produce the Visual Arrangement Plan for it. 221 School of Humanities and Social Science Division of Humanities Eccentricity Behind Beauty: Psychopathology and Artistic Expression in Chinese Painting Supervisor: FU Li-tsui / HUMA Student: TONG Yvonne Chung Sze / GCS Course: UROP2100, Fall Over the past decades, the examination of psychopathology has evolved from an esoteric theme into a well-recognised, even common, approach to understanding stupendous works of art. As psychiatric analysis grew increasingly prevalent, rising from the breakthrough beyond iconographic interpretation, so did the enrichment for aesthetic appreciation among artistry and artwork, Regarded as one of the four Great Masters of the Yuan dynasty, the fourteenth-century Chinese literati painter Ni Zan (1301-1374) provides an excellent case study for exploring the intricate relationship between the psychiatric wellness of the artist and the development of his artistry and the creation of his artworks. Some art historians, such as Fong Wen, describe Ni Zan as “a humanistic manifestation of united conflicts; an individual with both tameness and absurdity”. Such assertion, however, was not adequately explored and justified through detailed inspections of his paintings. On this premise, this research seeks to focus on uncovering the significance of psychological state on influencing and paralleling the artistic imagery and pictorial depiction of reality as presented in the paintings by Ni Zan. In placing the psychiatric wellness of Ni Zan the centre of this research, as well as taking into consideration his familial circumstances and the historical background of dynastic rivalry between the Yuan and Ming administrations, this research attempts to bring a deeper meaning to Ni Zan’s paintings as not only mere symbolism for Chinese literati paintings, but also manifestations of the influence of psychosis on artistic expressiveness in a Chinese artistic and cultural context. 222 School of Humanities and Social Science Division of Social Science Division of Social Science AI for Social Innovation Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: CHAN Wan Ting / ECOF LI Wanshan / FINA NG Lok Lam / SENG TSANG Chin Yung Virginia / IPO Course: UROP1000, Summer UROP1000, Summer UROP1000, Summer UROP1000, Summer In this report, we include 4 case studies of social innovation products (Safecity, Calliso, Spot, AllVoices) which are AI-incorporated to tackle the problem of sexual harassment. We first introduce AI and social innovation and highlight the issue of sexual harassment. Then, we evaluate each product from different aspects, such as privacy protection, user-friendliness and accessibility, with the goal of providing useful insights for us to build a chatbot in fighting sexual harassment on campus. Generations and Gender Survey Work Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: LAI Man Ting / RMBI NG Kai Yuen Dominic / QSA Course: UROP1100, Fall UROP1100, Fall The GGS is an international survey instrument and is part of the Generations and Gender Programme coordinated by the Netherlands Interdisciplinary Demographic Institute (NIDI). It collects information on inter-generational relations, gender roles and work culture. As a UG student, our duties were mainly focusing on tasks related to the pilot study such as translation, localization and works related to the beta testing. Not only does GGS allow us to acquire knowledge about social topics, but it also strengthens our critical thinking, problem-solving and communication skills. We have also cultivated to be more patient and detailed-minded. We have become more focused on the organization and coherence of questions and the structure of the survey. 223 School of Humanities and Social Science Division of Social Science Generations and Gender Survey Work Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: LAI Man Ting / RMBI NG Kai Yuen Dominic / QSA Course: UROP2100, Spring UROP2100, Spring We were involved in the conduct of the Hong Kong Generation and Gender Survey (HK-GGS). The HKGGS collects information of different generations, where the questions are related to daily lives such as interpersonal relationships, life satisfaction, gender roles and working environment. Our tasks were mainly focused on the preparatory activities – e.g., translating and localizing the survey questions, proofreading, beta testing and fixing errors of the online questionnaire, and preparing the invitation and reminder letters. Our involvement in these activities have improved our soft skills such as organization, communication and critical thinking skills. Some of the tasks required interacting with the other team members which greatly improved our interpersonal skills. We also gained a deeper understanding of how to run an online social survey. Our teamwork spirit has also been fostered through this project. Machine Learning for Processing Historical Images Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: SRIRAM Shriya / SENG Course: UROP1000, Summer Population figures can act as indicators for the living standards of people in certain areas at specific times. It reveals information about and observations on the differences among various groups in a society including but not limited to ethnicity, gender, and occupation. These differences can be effectively quantified and observed in the analysis of population statistics. In this project, we hoped to place our focus on the past and explore any important trends in the records of births and deaths that may inform us about potential inequalities faced by different population groups. Data on births and deaths in Wilmington, New Castle, United States from the late 1800s to the early 1900s was interpreted, transcribed, and studied. The most significant findings highlight that the birth and death rates may have been significantly impacted by social situations on race and the economy. 224 School of Humanities and Social Science Division of Social Science Reconstructing Historical Lives from Demographic Data Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: AGARWAL Aryawardhan / SENG Course: UROP1000, Summer As a group, we were tasked to input data from Wilmington, Delaware in the late 1800’s to early 1900’s. This data was taken from scanned records using FamilySearch’s services, where we looked at births and deaths in the region. Since the records were physical copies, we were tasked to interpret the records, then inputting the data into an excel sheet to make for easier analysis. The purpose of this data was to be able to identify trends based on different variables, like race, sex, location, and more, in relation to the births/deaths. These trends could then be compared to norm for that time. For example, one area we could look at is infant mortality rate, based on sex, to see whether female or male babies tend to survive longer. Similarly, we could look at average of death between the sexes and then conclude reasoning behind it based on the era’s treatment of women compared to men. TASKS OVERVIEW Population figures, in many cases, can act as indicators of the living standards of people in a certain area and time. It reveals information and observations on the differences among various groups in a society: ethnicity, genders, occupations, and more. These differences, often being included in public policy discussion, can be effectively quantified, and observed in the analysis of such population statistics. In this project, we hope to place our focus on the past and to explore if there were important trends in records of birth and death that may inform the characteristics on potential inequalities faced by different population groups. Data on births and deaths in the late 19th century and early 20th century is being studied, with a dataset on these records in New Castle, Wilmington, Delaware, United States. It is worthy of notice that there were already several studies focusing on related topics on population for the United States, such as mortality, but many were estimation-based and without details such that further analysis at other aspects can be done. Reconstructing Historical Lives from Demographic Data Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: HO Chi Hong / COMP Course: UROP1000, Summer As a group, we were tasked to input data from Wilmington, Delaware in the late 1800s to early 1900s. This data was taken from scanned records using FamilySearch’s services, where we looked at births and deaths in the region. Since the records were physical copies, we were tasked to interpret the records, then inputting the data into an excel sheet to make for easier analysis. The purpose of this data was to be able to identify trends based on different variables, like race, sex, location, and more, in relation to the births/deaths. These trends could then be compared to the norm for that time. For example, one area we could look at is the infant mortality rate, based on sex, to see whether female or male babies tend to survive longer. Similarly, we could look at the average death between the sexes and then conclude the reasoning behind it based on the era’s treatment of women compared to men. 225 School of Humanities and Social Science Division of Social Science Reconstructing Historical Lives from Demographic Data Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: LIU Junde / QSA Course: UROP1000, Summer Data Linkage is a hefty operation that is both time consuming and, more importantly, drains a ton of computational power. It involves running a matching-algorithm over every single possible combinations of records and determining whether or not there is a match. The data linking operation detailed in this report is completed by an AWS cloud server with 96 CPU cores running fully-loaded for 10 hours. Althrough the price to pay is high, the potential rewards for data linkage is also high. For a start, a well-linked dataset give the researcher more variables to work with. In our case, a linked dataset would allow us to look at the birth side and the death side of the equation together, instead of separately. Also, linking the datasets is the ultimate test to see if there’s any error present in the datasets. If the match rate falls short to our expectation, the only explanation would be that there are significantly amount of human error in the transcribing process. With this knowledge, we can then discover what went wrong in the transcribing process and then devise strategies to address the issues. Reconstructing Historical Lives from Demographic Data Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: MAHBUBANI Esha / SBM Course: UROP1000, Summer As a group, we were tasked to input data from Wilmington, Delaware in the late 1800s to early 1900s. This data was taken from scanned records using FamilySearch’s services, where we looked at births and deaths in the region. Since the records were physical copies, we were tasked to interpret the records, then inputting the data into an excel sheet to make for easier analysis. The purpose of this data was to be able to identify trends based on different variables, like race, sex, location, and more, in relation to the births/deaths. These trends could then be compared to the norm for that time. For example, one area we could look at is the infant mortality rate, based on sex, to see whether female or male babies tend to survive longer. Similarly, we could look at the average death between the sexes and then conclude the reasoning behind it based on the era’s treatment of women compared to men. Reconstructing Historical Lives from Demographic Data Supervisor: GIETEL-BASTEN Stuart Arthur / SOSC Student: WANG Xiaohan / GCS Course: UROP1000, Summer As a group, we were tasked to input data from Wilmington, Delaware in the late 1800s to early 1900s. This data was taken from scanned records using FamilySearch’s services, where we looked at births and deaths in the region. Since the records were physical copies, we were tasked to interpret the records, then inputting the data into an excel sheet to make for easier analysis. The purpose of this data was to be able to identify trends based on different variablessuch as race, sex, and location in relation to the births/deaths. These trends could then be compared with the norm at that time for further analyses. 226 School of Humanities and Social Science Division of Social Science Chinese and English Media Representation of Migrant Domestic Workers in Hong Kong Supervisor: GROVES Julian Mcallister / SOSC Student: WONG Mei Yue / CIVL Course: UROP1100, Fall The project examines migrant domestic worker-related articles on print media as well as documentaries aired on public channels in Hong Kong in 2014 and discusses how workers are represented differently. The student is involved in translating documentaries and quantifying the occurrences of several predefined music, visual images and interviewee perspectives of the documentaries using the software NVivo. Comparisons of documentaries produced in English and Chinese by the same media company are made. The data collected verifies the hypothesis that the Chinese media display signs of tabloidization and adopts an objectivity ritual, also selecting their perspectives featured to cater to its target audience, while the English media often stand with the welfare and rights of workers. Understanding Socioeconomic Determinants of Crimes in China Supervisor: HE Guojun / SOSC Student: CEN Yonglin / IS Course: UROP2100, Fall Violence against healthcare staff including doctors and nurses has been a long existing issue in China. These violent incidents often occurred as the offenders was trying to air their grievances and dissatisfaction. This study sets out to explore the determinants of such incidents by investigating the causes of patients’ dissatisfaction in healthcare settings. We have collected negative reviews on one of the largest healthcare community platform in China, Haodf, and will look into these comments and explore the major causes of patients’ dissatisfaction. This report summarizes the progress we have made and also reports the insights we have gained so far. Understanding Socioeconomic Determinants of Crimes in China Supervisor: HE Guojun / SOSC Student: WANG Yilin / GCS Course: UROP4100, Fall Back in the 1990s, there were several cases where the offenders killed the victims in coal mines and pretended to be their relatives to ask for compensation. The revealing of this type of crime that involves both murder and fraud shocked the public and it was named the blind shaft style crime because of a movie that reflects it. The egregious nature of such cases has caused concern in the community. However, due to the lack of research data, the literature on this type of crime is limited. This study regenerates data from previous research, judgment database and online news, aiming to provide an overview of this type of crime and explore the socioeconomic determinants of this type of crime. 227 School of Humanities and Social Science Division of Social Science Understanding Socioeconomic Determinants of Crimes in China Supervisor: HE Guojun / SOSC Student: CEN Yonglin / IS Course: UROP3100, Spring This study sets out to explore the socioeconomic determinants of healthcare violence in China. We looked into the incidents that have been reported by the media in recent years to understand the common characteristics of these incidents. We also collected patients’ negative reviews from one of the largest online healthcare community platforms in China to understand patients’ complaints. We summarized the data and found similar conclusions with previous studies. We plan to continue the study by investigating the news dataset with the patient review. By understanding the causes of the incidents, we hope to provide valuable insights for healthcare institutions and policymakers in China to develop effective strategies for the industry. Repressing the Emotion: Censorship of Popular Culture under Authoritarianism Supervisor: HONG Jean (Ji Yeon) / SOSC Student: FANG Zihang / GCS Course: UROP1100, Spring This Undergraduate Research Opportunity Program 1100C progress report aims to describe the general background of authoritarianism censorship in the milieu of People’s Republic of China and the preliminary assignments done during the first session of the Program. Since the program is in progress, this report serves to report all the previous works which mainly cascade into two parts, literature review and descriptive data. For literature review, it mainly introduces two influential scholarships on authoritarian censorship recently and detailed description of the film censorship in China. For descriptive data, it introduces how the data is collected, followed by the comprehensive descriptive analysis of the data collected. It depicts the film production by month and year in 2020. It presents the relationship of geographic distribution of the films produced and the films censored. The future work will focus on the in-depth analysis of the film production before and after the outbreak of pandemic, and the rationale behind censoring films in China by investigating the relationship between thematic texts and chances of being censored, which is designed to help the CCP indoctrinate its population by filtering the culture products up to the approved moral ideologies and official descriptions. 228 School of Humanities and Social Science Division of Social Science Repressing the Emotion: Censorship of Popular Culture under Authoritarianism Supervisor: HONG Jean (Ji Yeon) / SOSC Student: FANG Zihang / GCS Course: UROP2100, Summer The Undergraduate Research Opportunity Program 2100D report aims to decipher the text data based upon the summary of films’ abstract from the China Film Administration by means of the R’s package, Quanteda to build pertinent text analysis. The report will concentrate on two parts: one is the approach to text data coding; the other one concerning the frequency and textual analysis of the results by both virtualization and tabulation. The study may contribute a humble part to the subtle comprehension of the CCP’s censorship commitment to the popular culture products, especially to the film industry which is constantly imposed upon strict disciplines by film production regulations for indoctrination, promulgation of socialist positive mainstream and constrict products implicitly or explicitly conveying ideas disapproved of by the authority and socialistic code. Changing Alliances in Trump’s White House: Networks and Elite Politics Supervisor: KELLER Franziska / SOSC Student: CHENG Yu / GCS Course: UROP2100, Fall This UROP project is part of the “Changing Alliances in Trump’s White House: Networks and Elite Politics” research project conducted under the guidance of Professor Franziska Keller of SHSS. The goal of this Series 2 UROP project is targeting the U.S. Senate Intelligence Committee's Russian Active Measures Campaigns And Interference The 2016 U.S. Election Report’s Volume 5: Counterintelligence threats and vulnerabilities. An analysis of the behaviours, relationships and social networks among the people and institutions that were identified in this investigation was conducted under the project. To exemplify this effort, this report selects 1105 relationships that were identified from the Senate Report, analyses these relationships quantitatively, and confirms these results by qualitatively sorting out the formation mechanism of these relationships especially those of statistics focus. Changing Alliances in Trump’s White House: Networks and Elite Politics Supervisor: KELLER Franziska / SOSC Student: NI Yushang / GCS Course: UROP1100, Spring The report addresses political actors, political opinions, and communications as analytical elements. It then describes that during the process, the student utilizes social network analysis theory and finds out that natural language processing and graph illustration are two most adoptable strategies for investigating, predicting, and identifying politic-related affairs. The report also displays how the student creates directive graphs based on text and translate connections to concrete numbers. The visualization helps to predict stances, divergence, and cooperation in future political circle. Furthermore, it shows that the program is important to students due to its close linkage with the undergoing political and social events as well as the language analysis techniques which can be extensively used in future broaden research fields. 229 School of Humanities and Social Science Division of Social Science Changing Alliances in Trump’s White House: Networks and Elite Politics Supervisor: KELLER Franziska / SOSC Student: UPADHYAY Ritika / FINA Course: UROP1000, Summer This report describes the research project that aims to understand the alliances in Trump's White House during the period of the 2016 United States Elections. The data for the analysis is collected from the 116th Congress Senate Report and is used to identify networks among people to better understand the relationships among relevant political actors. A social network analysis tool by the name of Gephi was used to examine these networks in greater detail. This report discusses the findings, issues and conclusions derived from this examination. Political Astroturfing Worldwide: How to Manipulate Public Opinion on Twitter Supervisor: KELLER Franziska / SOSC Student: CHAN Ho Ting / QSA Course: UROP1100, Fall Twitter periodically releases data sets on suspicious accounts related to alleged political astroturfing campaign. With the release of a set of data related to a group of patriotic Cubans, this report found several abnormal characteristics presented in the group, these included a sudden increase of account creation and activity when there was a simultaneous drop of internet penetration in Cuba, highly active Twitter behaviour from the group, inconsistency between account language and tweet language, suspicious popularity of certain Twitter accounts, and that some of the users reported location correlates with a much higher Twitter activity comparing with the other users. By discovering these abnormal characteristics of the group, this report aims to help develop methods for identifying political astroturfing campaigns and discuss future investigation directions. Political Astroturfing Worldwide: How to Manipulate Public Opinion on Twitter Supervisor: KELLER Franziska / SOSC Student: KIM Hanbin / BIBU Course: UROP1100, Fall This report analyzed the tweets and retweets of 70 twitter accounts who are attributed to state backed information operations originating from Ghana/Nigeria targeting the 2021 US election. From the graph plotting the number of tweets or retweets against years, a sudden increase in activity is observed during the second half of 2019. Hence, this report divides into two sections: before and after 2019 January for the analysis. In the analysis, the number of accounts created before and after 2019, the number of followers for each account, hashtag trend and word cloud were used to highlight how this political astroturfing campaign took place. This report also further gave a chance to think about the outsourced political astroturfing campaign around the world. 230 School of Humanities and Social Science Division of Social Science Political Astroturfing Worldwide: How to Manipulate Public Opinion on Twitter Supervisor: KELLER Franziska / SOSC Student: CHAN Sum Yee / QSA Course: UROP1000, Summer Twitter political astroturfing means the use of automated accounts, known as “social bots”, “cyborgs” and accounts run by actual human to tweet, retweet and post videos to influence public opinions and political stance, with the aim to acquire certain political outcomes. Twitter political astroturfing is common in many countries and over the years, social scientists have been trying to find out the characteristics and patterns of astroturfing accounts for the easier identification of disinformation campaign, which helps to crack down on activities that are detrimental to democracy. Previous research has discovered patterns of astroturfing accounts affiliated with the Chinese government (Schliebs, Bailey, Bright and Howard, 2021) and the South Korean government (Keller, Schoch, Stier and Yang, 2019). In this report, I am going to investigate in the characteristics and patterns of suspected twitter accounts affiliated with the Russian government. Political Astroturfing Worldwide: How to Manipulate Public Opinion on Twitter Supervisor: KELLER Franziska / SOSC Student: LAM Ka Yan / GCS Course: UROP1000, Summer Since individuals are prone to adopting the opinions of most people, governments across the globe have launched various astroturfing campaigns to deceive the public into believing that their opinion or position is shared by the majority. By political astroturfing, participants are likely to pretend to be ordinary citizens acting independently in order to influence electoral outcomes and other forms of political behavior. However, without enough data, it is neither convincing nor persuasive to evaluate the occurrence and effectiveness of political astroturfing. As a result, it is of most importance for us to identify the information providers and information spreaders in various political astroturfing campaigns. Political Astroturfing Worldwide: How to Manipulate Public Opinion on Twitter Supervisor: KELLER Franziska / SOSC Student: WANG Yiren / QSA Course: UROP1000, Summer Political astroturfing is a centrally coordinated disinformation campaign in which participants pretend to be ordinary citizens acting independently, with the intention to influence the public voice. This study will identify and examine suspicious accounts among the followers of the Chinese diplomatic ambassador's Twitter account. By using various tools of RStudio to analyze the account creation dates, number of tweets, language, content and time of the last tweet of the suspicious users, the results showed that a large number of accounts were created on March 17, 2020, which further narrowed down the list of suspects. 231 School of Humanities and Social Science Division of Social Science Political Astroturfing Worldwide: How to Manipulate Public Opinion on Twitter Supervisor: KELLER Franziska / SOSC Student: MOON Kyoungin / QSA Course: UROP1100, Summer This project investigates the South Korean Embassies’ Twitter accounts and other related accounts in an attempt to trace any suspicious patterns that might indicate the presence of political astroturfing. Political astroturfing is often used by political power in an attempt to influence public opinion on social media by mimicking ordinary users. This report looks at when the embassy accounts were created and their activity status, what the accounts are talking about, and who they might be influencing to identify any suspicious patterns using the dataset provided by Twitter. All findings are analyzed through R studio and put into different charts and tables. State Visits and International Get-Togethers: Studying the Meetings of State Leaders Worldwide Supervisor: KELLER Franziska / SOSC Student: LI Jiting / MAEC QIAO Tianyi / GCS YEUNG Wing Ching Vinnci / IS Course: UROP1100, Fall UROP1100, Fall UROP2100, Fall When it comes to state visits, the media tend to sensationalize the influence and significance of the meeting, believing that state visits indicate turning points or major milestones in bilateral relations. Yet, little was known in terms of the “unspoken” rules behind each state visit, like who should make the visit and who should receive the visit, where should the visit take place, or what do we expect from each state meeting (Keller,2019). Thus, this project seeks to record and gather ground truth data that will help with the study of general social rules behind state meetings by utilizing data from Wikipedia, Nexis Uni and other mass media. We will also come up with primary assumptions and conclusions on this topic for future references. The Socioeconomic Impacts of Environmental Issues in Hong Kong Supervisor: LIU Tong / SOSC Student: LEUNG Ho Long / MAEC Course: UROP1100, Fall Pre-existing studies have not reached consent on the effects of public transport on the housing market (e.g., Mulley and Tsai, 2015; Zhang and Liu, 2015). Moreover, most of them only consider the direct effect of increasing convenience but not the indirect health risk caused by air pollution from public transport (e.g., Tian et al. 2017). This study aims to understand both the capitalization effect of increasing bus accessibility and the effect of air pollution created by buses. The project is in progress, and the cleaning of housing data is partly completed. With the previous bus stop data and air quality data, it is expected to combine all three in the future to estimate the willingness to pay for bus accessibility and minimizing health risks. Hedonic pricing and difference-in-differences will be performed to compare houses within a certain range and provide a well-round analysis of the benefits and costs of public transportation. 232 School of Humanities and Social Science Division of Social Science The Socioeconomic Impacts of Environmental Issues in Hong Kong Supervisor: LIU Tong / SOSC Student: LEUNG Ho Long / MAEC Course: UROP2100, Spring Following the previous UROP 1000 and 1100, the data cleaning is completed at the current stage. It consists of explanatory variables of housing prices, bus stops and air pollutants. Hedonic pricing and fixed effects with clustered standard errors are used to examine the relationship between these three. The presence of mixed results implies that we need further investigation before proceeding to the conclusion. Therefore, in the future, we would also take the number of floors into account to analyse the pollution effect on housing price. Aggregating the data at the monthly level is also needed to cope with the adjustment time in response to the changes in bus stops. The Socioeconomic Impacts of Environmental Issues in Hong Kong Supervisor: LIU Tong / SOSC Student: LEUNG Ho Long / MAEC Course: UROP3100, Summer Following the previous UROP, the data cleaning is completed at the current stage. It consists of explanatory variables of housing prices, bus stops and air pollutants. Estate and year-by-month fixed effects with clustered standard errors are used to examine the relationship between these three. Results show that the effect of bus stop to housing price is insignificant. Interestingly, the results change if a smaller sample is used. Belt and Road Initiative Supervisor: PARK Albert Francis / SOSC Co-supervisor: TRITTO Angela / IEMS Student: XU Anqi / MAEC Course: UROP1100, Summer Since its launch in 2013, China’s Belt and Road Initiative (BRI) has enabled the Chinese government to actively invest in nearly 170 countries globally with a focus on large-scale infrastructure construction projects and energy goods. This progress report presents how the Belt and Road Initiative affects China’s different types of trade and OFDI patterns at aggregate and sector level. We found that at aggregate level, countries receiving higher amount of OFDI from China also have high level of trade flow with China. Also, being a initially targeted BRI country will increase the possibility of receiving higher OFDI given the same amount of trade value with China. For energy sector, there also exists positive association between trade flow value and China’s OFDI, and being a BRI country will increase its OFDI given the same amount of China’s import to that country. 233 School of Humanities and Social Science Division of Social Science Greater Bay Area Supervisor: PATCHELL Gerald Roe / SOSC Student: LI Xinyue / GCS Course: UROP1100, Fall To promote the development of Guangdong, Hong Kong and Macao Bay area is a major decision and deployment made by the CPC Central Committee and the State Council to actively adapt to the new economic development. According to the development vision of Guangdong-Hong Kong-Macau Greater Bay Area (GBA), the ecological and environmental problems of GBA are still prominent; based on the strategic positioning of world-class urban agglomeration, GBA faces challenges in leading green development. The government should firmly implement the concept of 'ecological priority and green development', accelerate the improvement of the top-level design of ecological environment protection, continuously deepen the cooperation of ecological environment protection among Guangdong, Hong Kong and Macao. In this project, I am mainly in charge of the environment development of GBA, focusing on the ecological protection and planning of GBA. In this article, I would like to demonstrate my research on the ecological development of Guangdong-Hong Kong-Macau Greater Bay Area, in regard to the formation of the Pearl River Delta, wetland resources, ocean ecosystem, blue and green corridors, and national forest city clusters. Greater Bay Area Supervisor: PATCHELL Gerald Roe / SOSC Student: KEUNG Yeuk Yi / GCS Course: UROP1000, Summer Social sustainability frequently correlated with urban renewal, as most studies showed that urban renewal would effectively improve social sustainability for local residents. Social sustainability studies with associated with urban renewal process has not been largely discussed for academic research. This study located in Nantou Ancient City in Shenzhen, a prosperous metropolis in Greater Bay Area in China, through fieldwork and secondary resource analysis, research will analyze how social sustainability generated in an urban village with more than 600 years of history. The research finding will provide an empirical evidence for other scholars and governments to further study about this topic, or have more comprehensive decisions in urban planning. China/Africa Links Project Supervisor: SAUTMAN Barry Victor / SOSC Student: WONG CHEUNG Kin Bong / GCS Course: UROP1000, Summer The purpose of this research paper is to investigate into the link of China and four Africa countries — Sierra Leone, Liberia, Gambia, and Guinea-Bissau — especially in terms of agricultural links. The methodology is utilizing the second-hand resourse and relevant data on different database and different news media and throughout internet. In recent years, those four countries both have increasingly tied links with China in different aspects, ranging from mining sector to infrastructure, from agricultural sector to bilateral trade/investment. The influence that people perceived is of both positive and negative in different countries, in different aspects and in different stakeholders. 234 School of Humanities and Social Science Division of Social Science China/Africa Links Project Supervisor: SAUTMAN Barry Victor / SOSC Student: CHENG Yu / GCS Course: UROP1100, Summer This progress report is part of HKUST's Undergraduate Research Opportunity Program and is targeted on the China-Africa Links research, under the guidance and supervision of Professor Barry Sautman and Professor Yan Hairong. This report aims to summarise the contribution to the project over the semester, with an emphasis on the political and economic nature of China's relations with African countries, concentrating on the trade and investment connections among the bilateral relationships, the agricultural cooperation as another focus, and the evaluation of such connections by international public opinion. The research methodology is mainly a documentary review and data collection, drawing on data from the World Bank, WTO and specific governments to obtain an overview of the political and economic status in which each African countries stand, measuring the degree of economic proximity between China and that specific African country, referring to official treaties, documents and press reports to ascertain different patterns of agricultural cooperation between China and African countries, and finally collecting a wide range of media and academia perceptions to shed light on international public opinion. The results of these studies are differentiated by country and are presented to supervisors on a weekly basis. This progress report will unfold six countries involved in the said period’s study, to compare their socioeconomic development, the extent to which they maintain intense trade and investment relations with China, and typical examples of different modalities of agricultural cooperation. This comparative analysis is intended to capture which factors play an active role in China-Africa relations, and which aspects China-Africa cooperation finds promising or challenging. Framing 'Democracy' in Hong Kong Supervisor: WONG James K. / SOSC Student: CHAN Hei / MGMT Course: UROP1100, Fall In light of the fray over the social reality in Hong Kong, this study delves into the discourse between the blue (pro-establishment) camp and the yellow (pan-democratic) camp to examine the duality of interpretations in Hong Kong on issue framing and political notions, pertaining to democracy and the identity of the people of Hong Kong. It is observed that populist rhetoric is used in the political discourse of Hong Kong. The three core elements of populism function as the diagnostic, prognostic and motivational frame of the social reality exploited by the blue and the yellow camp respectively. The tendency towards populism introduces questions over the role of emotions and radicality in Hong Kong politics. 235 School of Humanities and Social Science Division of Social Science 236 Identifying Protest Events in China with Social Media Data Supervisor: ZHANG Han / SOSC Student: WANG Yiwei / QSA Course: UROP2100, Fall This UROP2100 report is the following report of my previous UROP1100 project: Identifying Protest Events in China with Social Media Data conducted under the supervision of Professor Zhang Han, School of Humanities and Social Science. The data of this study is from the 9.5 million Weibo (the Chinese Twitter) posts from 2011 to 2017 classified by the CASM (collective action from social media) system (Zhang & Pan, 2019). This project continued the data coding in the last UROP project and has completed some basic analysis on source type, issue type, events duration, and events by provinces of the coded data. Among the 1,001 coded events, 91.6% of posts are from Weibo, issues related to minorities, taxi driver, and commercial fraud are most likely to become protests. The mean for the duration of events is 54.8 days, and the median is 4 days. Among all the provinces, “Guangdong'', “Sichuan”, and “Shandong” have the highest probability of events happening while there are no events in Tibet. Representation Learning of Social Survey Data Supervisor: ZHANG Han / SOSC Student: LYU Hanfang / MATH-AM Course: UROP1100, Spring For many years, social scientists have used surveys as a research technique. Quantitative analysis of survey data from a social science viewpoint has gained increased interest as a result of the newly emerging trend of computational social science. While classical statistical machine learning strategies have been extensively employed in survey analysis, few state-of-the-art innovative machine learning approaches have been applied in this field of research. In this report, we employed graph learning methods, such as Graph Convolutional Networks (GCN), to accomplish a pre-defined survey analysis tasks, cross domain response prediction, and we also offered results from similar analysis by statistical machine learning methods. To that purpose, we may do a thorough comparison of the two methodologies, which will help us better understand future survey analysis using GCN. Representation Learning of Social Survey Data Supervisor: ZHANG Han / SOSC Student: ZHANG Hao / COMP Course: UROP1100, Summer We aim to use graph convolutional networks (GCN) to analyze social survey data. Now we are focusing on the missing data problem. We have tried to apply different models and methods on the data set: zuobiao1214_clean.dta to predict some missing data. But so far, we didn’t find a such good way that the accuracy of prediction can exceed 65%. In this report, I will elaborate on how we constructed the graph and which model we have tried. Abstracts of UROP Projects 2020-21 Interdisciplinary Programs Office Interdisciplinary Programs Office Division of Environment and Sustainability Division of Environment and Sustainability Fuel and Mobility Poverty in Hong Kong Supervisor: DELINA Laurence Laurencio / ENVR Student: PARK Jungjoon / EVMT Course: UROP1100, Fall This project developed a short research proposal for assessing the prevalence of fuel and transport poverty in Hong Kong. Poverty surveys done by the Hong Kong SAR is currently limited to how many households are lacking income. A recent trend of global poverty studies pays more attention to understanding the inaccessibility of goods necessary for households to satisfy their basic needs. This research proposal aims to measure the prevalence of fuel and transport poverty in Hong Kong’s North and Kwun Tong administrative districts through household surveys and spatial analysis using Geographic Information System. The final output will include key fuel and energy poverty statistics in the test districts. Interrogating Fintech-Enabled Solar Power as ‘Energy Democracy’ in Hong Kong Supervisor: DELINA Laurence Laurencio / ENVR Student: LAM Rainbow Yi Hung / EVMT Course: UROP1100, Fall This UROP seeks to examine the status and potentials of a fintech-enabled renewable energy market and its relationship with energy democracy in Hong Kong. During this term, we conducted a literature review, developed the first draft of our interview guide, identified potential interviewees/respondents, and completed our applications for research ethics. Our literature review suggests the following: (1) Blockchain can help improve cash flow of businesses that provide renewable energy; (2) Areas with developed financial markets are seen to have more growth in the renewable energy sector; and (3) Blockchain application is not widely seen in the oil and gas markets. Our report concludes with an outline of our future work. Interrogating Fintech-Enabled Solar Power as ‘Energy Democracy’ in Hong Kong Supervisor: DELINA Laurence Laurencio / ENVR Student: LAM Rainbow Yi Hung / EVMT Course: UROP2100, Spring This UROP seeks to examine the status and potentials of a fintech-enabled renewable energy market and its relationship with energy democracy in Hong Kong. During this term, we finalized our principal research instrument, an interview guide, and firmed up our list of interviewee/respondents. We also ran a pre-test of our interview guide and modified our interview protocol based on this trial run. Our pre-test suggests, preliminarily, (1) perception on the increased deployment of solar energy in Hong Kong, (2) barriers in terms of lengthy project approval process, limited land area, and solar panel inefficiency, and (3) less familiarity with concepts of “energy democracy” and “fintech-enabled solar power.” Our report concludes with an outline of our future work. 238 Interdisciplinary Programs Office Division of Environment and Sustainability Interrogating Fintech-Enabled Solar Power as ‘Energy Democracy’ in Hong Kong Supervisor: DELINA Laurence Laurencio / ENVR Student: TANG Wing Shun / EVMT WONG Ka Ying / EVMT LAM Rainbow Yi Hung / EVMT Course: UROP1000, Summer UROP1100, Summer UROP3100, Summer This UROP seeks to examine the status and potentials of a fintech-enabled renewable energy market and its relationship with energy democracy in Hong Kong. During this term, we conducted our primary data collection, through online interviewing. We also transcribed our data and cleaned our transcripts. Our initial findings are: (1) Majority of our resource persons think that solar energy and financial technologies are key instruments that can assist Hong Kong to transition towards carbon neutrality; (2) Majority of them also think that there are limitations and challenges that need to be addressed if Hong Kong is to tap this opportunity, including lack of expertise and lack of interest from the business sector. Our report concludes with a summary of the challenges we encountered in data collection and an outline of our future work. Air Quality Data Analysis and Interpretation Supervisor: GU Dasa / ENVR Student: WANG Yiyi / CHEM Course: UROP2100, Fall Based on my research done in UROP1100, combining with the characters of air pollution in Hong Kong, I found ozone of most interest of discussion and that NOx was one of the most important precursors of ozone. Hence I further dug into the correlation relationship between concentrations of ozone and NOx using air quality data acquired from HKEPD website and made some observations on diurnal patterns, weekday-weekend characters, monthly fluctuations and seasonal patterns of NOx and ozone concentrations in three places with diverse land-use characters in Hong Kong. Air Quality Data Analysis and Interpretation Supervisor: GU Dasa / ENVR Student: WANG Yiyi / CHEM Course: UROP3100, Spring Based on my fundamental research on influence factors of ground-level ozone in UROP1100 and UROP2100, which extracted HKEPD data on pollutants concentrations and HKO data on natural parameters, I found out that concentrations of ozone is more strongly correlated with concentrations of its precursors, VOCs and NOx, and through further readings, I discovered that VOCs have a more significant effect on ozone than NOx. In this report, I will focus on BTEX family of VOCs, study their diurnal and seasonal fluctuation patterns by plotting time series, and discuss the Pearson’s correlation coefficients between concentrations of BTEXs and ozone, adopting pollutants concentrations data of Mong Kok station, from HKEPD public site and monitoring data of our labs in HKUST, from May1 2015 to May31 2019. 239 Interdisciplinary Programs Office Division of Environment and Sustainability Air Quality Data Analysis and Interpretation Supervisor: GU Dasa / ENVR Student: WAH Monique Zoe / SSCI Course: UROP1000, Summer GEO-KOMPSAT-2A (GK2A) is a geostationary satellite located at 128.2°E. Chlorophyll-a concentration can be calculated from GK2A data following Murakami’s method illustrated in the paper “Ocean Color Estimation by Himawari-8/AHI”. GK2A visible band data will first be displayed and merged as a true color image. GK2A’s Advanced Meteorological Imager (AMI) spectral response function will be plotted and analyzed for simulation of remote-sensing reflectance value. During the calculation process, Inherent Optical Property Model and Band-ratio Algorithm are utilized. NOMAD data is also used for deriving the coefficients of Band-ratio Algorithm. Chlorophyll-a concentration is displayed as a color map, and comparison between calculation result with MODIS and Himawari 8 product is present in the paper. Air Quality Data Analysis and Interpretation Supervisor: GU Dasa / ENVR Student: WANG Yiyi / CHEM Course: UROP4100, Summer I reported on the sensitivity of ground surface ozone to the changes on the fluctuations in concentrations of volatile organic compound (VOC) and nitrogen oxides (NOx = NO + NO2) observed in a windless dry sunny day at the coast of Silver Strand Bay, using a 0-D box model which has the Master Chemical Mechanism implemented for simulation as well as Python for time series plotting and correlation analysis. The findings of this study educated me about VOCs and their impact on O3 formation, and will be helpful in formulating emission control strategies for coping with O3 pollution related to photochemical air pollution in Hong Kong. Use WiFi Technologies to Predict Building Energy Consumption Supervisor: LU Zhongming / ENVR Co-supervisor: QU Huamin / CSE Student: CHAWLA Anhad Singh / SUEE Course: UROP1100, Fall This research report explores the relation between several variables that predict energy usage by creating machine learning regression models. Data for new variables regarding building use type was collected using Google Maps and Street View and added to the existing database. The code was adjusted to accommodate the new data and after some data cleaning, 13 models used the data to create a predictive model. The specially curated database is discussed in detail with an explanation for the variables chosen for model input. The efficacy of the models is discussed and future research directions are suggested. 240 Interdisciplinary Programs Office Division of Environment and Sustainability Visualization of Tree Failure in Super Typhoon Mangkhut Supervisor: LU Zhongming / ENVR Student: WAHYU Axell Nathaniel / CIVL Course: UROP1000, Summer This project tries to figure out a better urban design that can reduce the risk of having trees falling over by analyzing the tree failures that were caused by Typhoon Mangkhut in 2018 in Hong Kong. With fewer collapsed trees, there will be less property and consequently monetary damages while making a safer environment for the public during extreme weather phenomena. Currently, the project has utilized geographic information system software (ArcGIS) to help visualize the tree failures over Hong Kong for easier risk and design analyses. The communication with Hong Kong’s Development Bureau is still going to obtain a set of data of trees that are currently standing in the city, which is necessary before being able to properly calculate the risk. Several potentially influencing factors have been made which includes the wind speed, soil strength, sky view factor, and urban heat island. The future direction of this research project is guided to some extent by understanding where these factors come from and what they can do towards the trees in an urban setting. Ultra Compact Air Sensor Development for Personal Health and Indoor Air Quality Supervisor: NING Zhi / ENVR Co-supervisor: LAU Alexis Kai Hon / ENVR Student: RATTANAMANEEJARAT Thananun / SENG Course: UROP1000, Summer Air pollution is a field of growing concern, and bacterial activity in the air in one aspect of this. As bacteria can produce lactate through the process of anaerobic respiration (up to 600 mM), lactate sensors are a candidate for the detection of bacteria in air samples. However, as each biosensor has a linear range of measurement of analyte concentrations, and bacteria produces more lactate than that range, a diffusion layer (a semi-permeable membrane controlling the diffusion of the analyte) is placed atop the enzymatic region to extend the linear range. The primary aim of this report is to understand the relevant factors in optimizing and using a diffusion layer on an amperometric lactate enzyme sensor. 241 Interdisciplinary Programs Office Division of Environment and Sustainability Affordable, Sustainable Fashion!? Supervisor: SAUERWEIN Meike / ENVR Co-supervisor: MATUS Kira / ENVR Student: AGCAL Bengi / CPEG DATTA Sharnasree / SBM SUTEDJA Amrita Saraswati / ACCT YA HUI XIAO Julia / BIOT Course: UROP1000, Summer UROP1000, Summer UROP1100, Summer UROP1100, Summer Covid-19 is an unprecedented event that led to drastic changes in people’s lives, including consumption patterns. As people have less opportunities to go out and meet others, the need to consume new or “fashionable” apparels may change too. This study aims to analyze the impact of the pandemic on HKUST student and staffs’ apparel consumption behavior by distributing a survey. The key findings from the survey are the majority of respondents did not change their consumption behavior during the pandemic; minimalism is seeing an uptrend but it is inconclusive whether the cause was sustainability awareness, and online shopping becames a new trend that may have caused people to maintain their apparel consumption intensity. Further research and interviews are needed to determine the cause-effect relationships of the pandemic, change in income and behavioral patterns, and apparel consumption behaviors in Hong Kong. Apps and Tools to Encourage Responsible, Sustainable Consumption Supervisor: SAUERWEIN Meike / ENVR Student: YU Wei Lin Whitney / EVMT Course: UROP2100, Fall Sustainable consumption is inherently tied to the choices we make in our daily lives, but very rarely are students aware of the direct effects their decisions have on the environment around them. This UROP project aims to develop a platform and toolkit for teachers, providing materials and guidance on teaching sustainable consumption in Hong Kong. To accomplish this, we developed prototype class materials for secondary school teaching, to be tested by teachers and their students in a workshop setting and evaluated the materials’ suitability for high school teaching and ability to engage participants in the topic of sustainable consumption. Further research will be conducted through a survey developed to collect teachers’ first-person experiences regarding sustainability education and teaching resources usage. Apps and Tools to Encourage Responsible, Sustainable Consumption Supervisor: SAUERWEIN Meike / ENVR Student: YU Wei Lin Whitney / EVMT Course: UROP3100, Spring Despite the increasing relevance of sustainability in all sectors and walks of life, students of Hong Kong are rarely aware of the environmental impacts of their decisions in daily life. This project aims to analyse the educational needs of educators with regards to sustainability, in order to determine how to effectively address the gap in students’ understanding of sustainable consumption. Through educator engagement and research via a workshop, survey, and interviews, it is readily apparent that despite a general recognition of sustainability-related needs, the current pool of educational resources on sustainability requires localisation to become relevant in a HK context as well as adequate support in finding a niche within current syllabi. 242 Interdisciplinary Programs Office Division of Environment and Sustainability Apps and Tools to Encourage Responsible, Sustainable Consumption Supervisor: SAUERWEIN Meike / ENVR Student: GOSWAMI Arnab / CEEV LIN Yu-chun / SENG YU Wei Lin Whitney / EVMT Course: UROP1000, Summer UROP1000, Summer UROP4100, Summer Despite general awareness and concern for environmental problems, current students still fail to relate their daily life choices with sustainability consequences. This study aims to evaluate the effectiveness of various tools to encourage broader sustainable consumption education, namely through the study of using a serious game using surveys combined with facilitated test-play sessions, and by developing teacher resources to collect educator feedback. Collected feedback indicated that interactive tools such as serious games can raise students’ inclination towards sustainable consumption behaviours and teachers would be interested in such materials, thus demonstrating that given adequate incentive, resources such as lesson plans and serious games can contribute significantly to improving students’ understanding of sustainable consumption. Deceptive Environmental Claims and Greenwashing Supervisor: SAUERWEIN Meike / ENVR Student: FU Jiayi / EVMT SUTEDJA Amrita Saraswati / ACCT Course: UROP1100, Fall UROP1100, Fall Greenwashing has become more and more common on the market since the consumers pay more attention to sustainability and the ‘claim to green’ somehow attract the buyers and boost the sales volume. During the purchasing process, some consumers may or may not notice the eco-claims, which is hard to decide the authenticity, others may or may not care whether the promises are true, which leads to different purchasing behaviors. This project is designed to investigate consumers’ knowledge of concept and identification of ‘green’ and ‘eco-labels’, attitude and behavior towards green claims. At this stage, we are in the preparatory phase for the survey designed mainly for people with the decision power and usage rights to determine what to buy in grocery. The main focus is the design of the survey questions, including the question types, choice of words and phrases, as well as the order of questions. This report firstly discusses the key points for the survey design. Then, with the help of the research, we further update the original survey according to some rules and suggestions. Finally, some issues on the translation process are proposed and discussed. 243 Interdisciplinary Programs Office Division of Environment and Sustainability Deceptive Environmental Claims and Greenwashing Supervisor: SAUERWEIN Meike / ENVR Student: SUTEDJA Amrita Saraswati / ACCT Course: UROP2100, Spring With increasing amounts of Hong Kong retailers offering “green, sustainable products” on their shelves, the variety of ecolabels and marketing claims advertising sustainability attributes have blossomed. This is not only good news for consumers. It challenges consumers to differentiate certified eco-labels different countries from simple green-marketing symbols and to evaluate which sustainability claims are genuine and what is just greenwashing. To evaluate consumers attitude towards sustainable products and their ability to identify green washing, we conducted a survey among students and Indonesian domestic helpers, as well as follow-up interviews. Results show that both the students and helpers care most about price and quality/performance in buying green products, which could be a barrier to sustainable consumption in Hong Kong given the high price-premiums of such products. In terms of knowledge about product sustainability, students generally showed a better understanding of eco-labels and ability to differentiate greenwashing from genuine sustainability claims or fake eco-labels. Yet, when asked about their confidence/subjective knowledge towards being able to identify greenwashing, they rated their knowledge significantly lower compared to the score achieved when being tested on examples. The next step of this study would be to conduct offline surveys with the helpers to reach a larger group of respondents, and expand the target group to local households, combined with further follow-up interviews. Driving Behavioural Change of BYOBottle by Smart Data and Innovation (SSC Smart Meters Project) Supervisor: SAUERWEIN Meike / ENVR Student: OUDOMYING Por / SSCI CHAN Andrew Gachun / SBM Course: UROP1000, Summer UROP1100, Summer Given the increasingly imminent issue of climate change, the increase in use of single-use plastics given the COVID-19 pandemic, and the public’s misinformed hygiene concerns towards water refill stations, there is a need to introduce interventions to enhance the uptake of “Bring your own bottle” (BYOB) behavior. A multidisciplinary, comprehensive literature review has been conducted on the sociopsychological factors contributing to the consumption of single-use plastic bottles. Hence, a three-staged intervention relating to hygiene perceptions of refill stations, social impact visualization, and material rewards based on personalized tracking will be implemented in the coming year. With results studied with surveys and focus group interviews, we will study the efficacy of interventions based on various value frameworks under the distinctive COVID-19 context. 244 Interdisciplinary Programs Office Division of Environment and Sustainability Life Cycle Approaches for Sustainable Buildings and Infrastructures Supervisor: SAUERWEIN Meike / ENVR Co-supervisor: BOOKHART Davis Boyd / ENVR Student: CHUNG Kai Jing Jeremy / EVMT YEUNG Ho Fan Marco / EVMT Course: UROP1100, Fall UROP1100, Fall This research explores the impact of green product labelling schemes in the construction industry, and it’s relationship with Life Cycle Assessment (LCA) based approaches. The research focuses mainly on the Green Product Certification (GPC) by the Construction Industry Council (CIC), which is the dominant certification scheme today. Analysis is made on its assessment criteria, utilization of the LCA and the relations between the scheme and the Green Building standard BEAM Plus. Effectiveness and inclusiveness of the current labelling scheme are mentioned. Suggestions towards how the GPC and BEAM Plus could increase its popularity and utilization are provided. Life Cycle Approaches for Sustainable Buildings and Infrastructures Supervisor: SAUERWEIN Meike / ENVR Co-supervisor: BOOKHART Davis Boyd / ENVR Student: GOSWAMI Arnab / CEEV Course: UROP2100, Fall The LEED (Leadership in Energy and Environmental Design) and BEAM (Building Environmental Assessment Method) building standards are examined and a comparison is drawn in this report. The life-cycle analysis requirements for both standards are illustrated along with the Green Products declarations and their weightage in the overall points system is discussed. The report further studies the CIC Carbon Assessment Tool and the various categories of materials required to conduct a life-cycle analysis of a building. The depth of the Life-Cycle Assessment (LCA) specifications in the two building standards is also compared and areas of improvement are discussed. Extreme Weather under Global Warming Supervisor: SHI Xiaoming / ENVR Student: TAN Hanzhi / SSCI Course: UROP1000, Summer With a great passion for atmospheric science, I have been enrolled in this UROP1000 project. After that, I am more and more feeling that my ability in the researching is getting enhanced, in both Python programming and meteorology knowledge, thanks to the mentoring by my supervising faculty. However, with that progress, I think that I still have improving space for advanced research skills. In this report, I will describe the data analyzing method for the Coupled Model Intercomparison Project (CMIP) experiments, the knowledge I have learned about Mei-yu precipitation, and the connection between atmospheric circulation and Mei-yu rainfall in the CMIP simulations. 245 Abstracts of UROP Projects 2020-21 Institutes and Centers Institutes and Centers Institute for Emerging Market Studies Institute for Emerging Market Studies Trade and Investment under China's Belt and Road Initiative - Implications for Hong Kong Supervisor: TRITTO Angela / IEMS Co-supervisor: PARK Albert Francis / IEMS Student: WU Hongsen / GBUS Course: UROP1100, Fall Since the notion’s invention in 2013, Belt and Road Initiative (BRI) has become a centrepiece of China’s diplomatic policy. By encouraging Chinese capital to invest into infrastructure development in foreign countries, the BRI has extended Chinese economic and political influence in southeast Asia, central Asia and Africa. While there are many studies on the BRI, there is still very little written on what has this initiative created. Current empirical studies on how Chinese outward FDI have changed after the launch of the BRI show that investments have primarily gone into energy manufacturing and in building industrial parks and special economic zones. As part of this project, we are trying to build a unique series of dataset that show the amount of Chinese FDI that have happened into industrial parks and special economic zones, with special focus on 10 ASEAN countries. We can then utilize the datasets to analyse the effectiveness and efficiency of BRI and discuss the initiative’s impacts to both China and host countries. The project is still underway, and this progress report will discuss how we constructed the dataset of industrial parks for these countries, the methodology we used to match industrial parks with China’s FDI records, and the preliminary results we observed from Malaysia, Indonesia and Myanmar. Trade and Investment under China's Belt and Road Initiative - Implications for Hong Kong Supervisor: TRITTO Angela / IEMS Co-supervisor: PARK Albert Francis / IEMS Student: ZHANG Qingqi / ECOF Course: UROP1100, Fall This report summarizes my procedure of conducting analysis of key energy policies in all ASEAN countries during the fall semester. The specific tasks include but are not limited to outlining the renewable energy policies in the 10 ASEAN countries, constructing a brief dataset of energy production and consumption for 10 ASEAN countries using the BP Statistical Review of World Energy, as well as editing and updating the statistics and policies in two UOB reports concerning Belt and Road in Southeast Asia (Myanmar and Malaysia). 248 Institutes and Centers Institute for Emerging Market Studies Trade and Investment under China's Belt and Road Initiative - Implications for Hong Kong Supervisor: TRITTO Angela / IEMS Co-supervisor: PARK Albert Francis / IEMS Student: JIANG Yuxuan / ECOF Course: UROP1100, Spring Recent years have witnessed the significant effect of the Belt and Road Initiative on global economic and infrastructure development, and its impact on different fields has also become a topic of close attention. This project focuses on the energy sector and seeks to examine factors that motivate Chinese energy firms to invest in Southeast Asian countries under the Belt and Road Initiative. This report presents a preliminary selection of variables after the literature review process. The results show there are 5 dimensions with 22 potential push and pull factors affecting green and black energy investments, which are worth further data collection and regression analysis performance. Trade and Investment under China's Belt and Road Initiative - Implications for Hong Kong Supervisor: TRITTO Angela / IEMS Co-supervisor: PARK Albert Francis / IEMS Student: WU Hongsen / GBUS Course: UROP2100, Spring This progress report explains the objectives, methodology, and results of the construction of the first dataset that maps the flow of Chinese foreign direct investments (FDI) in Chinese overseas industrial parks in ASEAN countries. Through this dataset, we provide descriptive evidence explaining cross-country variation of FDI, as well as the role of state-owned enterprises provincial governments in facilitating such FDI. The results provide insights for further studies on the models and impacts of Chinese overseas industrial parks under the Belt and Road Initiative. Trade and Investment under China's Belt and Road Initiative - Implications for Hong Kong Supervisor: TRITTO Angela / IEMS Co-supervisor: PARK Albert Francis / IEMS Student: JIANG Yuxuan / ECOF Course: UROP1000, Summer Chinese medical aid and foreign direct investments are two main flows direction China’s current development, especially under the COVID-19 pandemic situation. With this large-scale foreign assistance and investments for decades, what attitudes the other countries take towards China gradually become a topic worthy of further investigation. This project gathered the voting records in United Nations Human Rights Council to reflect countries' attitudes in international organizations. Through the process of literature review, data collection and analysis of key variables, the researcher explored the effect of natural resources on voting decisions is not as significant as which on foreign direct investments, and the alignments inside different country blocs were robust. 249