Research Projects

Use deep learning-based numerical algorithms to solve partial differential equations that arise from science and engineering.

1) Have strong Mathematics background; familiar with programming; knowledge of numerical methods.
2) Prerequisites: multivariable calculus, linear algebra, differential equations, and numerical analysis.

Recent brain-wide calcium imaging techniques can record activity from the majority of neurons simultaneously in awake and behaving animals, such as zebrafish. This provides comprehensive information to explore the relationship between neural activity, brain functions, and behaviors. The project aims to adapt and improve machine learning methods to predict brain-wide neuronal activity using high-resolution behavioral videos recorded concurrently. This will allow us to isolate brain activity that is not directly related to motor actions or external stimuli, which could be associated with internal processing and brain states, and subject to the focus of further studies.

1) Read and reproduce existing data analysis methods in the literature.
2) Design and write computer programs for improved methods.
3) Apply the methods on experimental data, interpret the results and improve methods.

The student will investigate the coral-symbiont system, mainly conducting the culture and experimental evolution of the symbiotic algae Symbiodinium.

1) Conduct the culture and experimental evolution of the symbiotic algae Symbiodinium.
2) Learn basic skills of molecular biology.

The student will use developmental and molecular biology approaches to study the processes of the reverse development of the “immortal” Jellyfish Turritopsis sps.

1) Conduct development and molecular biology experiment related to the project.
2) Take care of and maintain the jellyfish in the lab.

The student will learn and experience basic molecular biology skills using non-model marine invertebrate systems such as jellyfish or corals.

1) Conduct basic research under guidance and supervision.
2) Take care of the marine invertebrates.

Study the evolution of marine invertebrates using bioinformatic tools.

1) Conduct research using bioinformatic approaches such as genomics and phylogenetics.

Warming of the ocean leads to thermal expansion of sea water, which has a contribution to sea level. The project aims to quantify the contributions to sea level rise in the South China Sea and Northern Pacific region using various data products, although the principal focus will be on the ECCO product. Data will be cross-calibrated and compared with existing tide gauge data in HK as appropriate. Experience with data analysis (ideally Python or MATLAB) and possession of a laptop is essential; experience with ocean science, sea level science or numerical modelling is desirable, but not strictly necessary.

1) Conduct literature review and learn of sea level science-related concepts.
2) Gather the relevant ECCO data.
3) Perform analysis of Ocean Heat Content within the South China Sea and Pacific region as appropriate (e.g., vertical distribution, changes in time etc.).
4) Convert the changes in Ocean Heat Content to thermosteric sea level rise.

Common chip manufacturing requires complicated equipment and facilities. Recently we brought a novel idea from basic physics principle and invented a novel laser writing technique that can deposit materials with high precision on a large range of substract. In this project, we will explore along the way for further development.

1) Develop new protocols and analyze the results.

Nitrogen vacancy center in diamond has long coherence time and can be controlled with high fidelity and high speed. In this project, the student will be involved in optimizing fidelity of quantum gates and increase the coherence time of the qubits.

1) Take and analyze data.
2) Be engaged in both experimental work and theoretical simulations.

Nitrogen vacancy center in diamond has became one of most powerful sensors for studying various physical phenomena. In this project, we will use it to study strongly correlated systems as they are the most interesting and challenging fields in physics ,including superconductivity and magnetism.

1) Help setting up the setups, take the data, analyze the data.

Many questions in biology are about the interaction of proteins. Manipulating force in biological systems can provide us some useful methods to address unaccessible questions using the current technique. We are building magnetic tweezers which can be used to study interactions between proteins and cellular process.

1) Build magnetic tweezers and perform pilot experiments with PG students under the guidance of a mentor.

Plastics are widely used all over the world and slowly degraded into nanoplastics. Airborne nanoplastics can be inhaled by humans and can cross cells in lungs and accumulate in tissues and organs, causing adverse effects in health. For example, nanoplastics can cause respiratory diseases including pulmonary infection and lung cancer. However, there is no effective way to prevent nanoplastics from accumulating or entering cells in lung. In this study, we will develop traditional Chinese medicine to clearing these nanoplastics in lung and prevent pulmonary infection.

1) Read the papers about uptake of nanoplastics into lung.
2) Perform the uptake assays of nanoplastics in lung cells.
3) Find the optimal traditional Chinese medicine to clear nanoplastics in lung cells.

By using single-particle cryo-EM, we aim to determine high-resolution structure of important membrane proteins, and to investigate their working mechanism by combining with biochemical, electrophysiological, and other approaches.

1) Work on protein expression and purification.
2) Prepare and evaluate EM sample.
3) Have basic knowledge on biological science, particularly protein biochemistry.

Our respiratory system is coated with motile cilia, tiny beat machines that work in synchronization to eliminate the invading pathogens. How these machines are built and organized is not clear. In this project, we will use CRISPR based genomic editing and super-resolution microscopy to study the protein-structure-function relationship, which is critical for the diagnosis of diseases caused by cilia defects.

1) Apply CRISPR genome editing tool to knock out genes governing cilia function.
2) Apply super-resolution imaging to observe cilia.

In this project, we aim to silence the cilia genes in cells to investigate their functions. Students are expected to develop the skills of CRISPR genome editing under supervision. He/she will learn all the procedures of CRISPR genome editing and contribute to the project by generating CRISPR KO cells. Subsidy may be provided depending on the performance.

1) Generate genome-perturbated cells under the supervision of a postgraduate student.

The student will be involved in the development and implementation of a high-throughput fluorescence imaging platform for long-term recording of single-cell dynamics in bacterial cells.

1) Prepare bacterial cultures for cloning and imaging experiments.
2) Perform plasmid library construction and quality control.
3) Perform PDMS microfluidic chip fabrication.
4) Perform quantitative fluorescence microscopy.

Based on the student's interest, different options could be available:
1) Building a microfluidic tumor on chip model for studying tumor microenvironment
2) Analysis and assembly of novel genomes from the rainforest to identify species and quantify biodiversity

1) Conduct computational analysis independently (with guidance from a PG student).
2) Use a microdevice for fabrication.
3) Assist with tumour-on-chip culture experiments.
4) Prepare progress reports bi-weekly.
5) Prepare a final presentation to the lab at the lab meeting at the end of the internship.

Synaptic transmission in Alzheimer's disease is a crucial question but remain elusive. In order to address this question, we will study how synaptic vesicles are released in neurons with amyloid. We will figure out how amyloid affects the exocytosis of synaptic vesicles.

1) Read the papers related to Alzheimer's disease.
2) Dissociate hippocampal neurons.
3) Perform the assay of synaptic vesicles in the presence of amyloid.

The apolipoprotein E (APOE) gene is the most significant genetic risk factor for Alzheimer's Disease (AD). Specifically, the APOE4 allele increases the risk of AD, while the APOE2 allele has a protective effect compared to the more common APOE3 allele. Apart from these apoE isoforms, there are additional variants, such as apoE-Christchurch and apoE-Jacksonville, which also show protective effects against AD, while others, such as apoE-Pittsburgh, are associated with increased AD risk. This project aims to analyse the biochemical and biophysical characteristics of these apoE variants, such as their aggregation and lipidation behaviours, to better understand their roles in AD pathogenesis.

1) Review literature to understand the project and identify knowledge gaps.
2) Support experimental design and protocol development.
3) Perform experiments, including data collection and analysis.
4) Collaborate on troubleshooting and discuss results.
5) Prepare reports and presentations to communicate findings.

Nanoplastics has become a complicated and urgent issue in the world. Nanoplastics can affect synaptic transmission in neurons. However, how nanoplastics affect synaptic transmission remains unknown. In this project, we study how nanoplastics affect synaptic transmission in living neurons.

1) Grow neurons.
2) Learn how to interpret data.

Nanoplastics become a complicated and urgent issue in the world. Nanoplastics can cause lung disease because they can be inhaled through lung and lead to lung inflation. However, how nanoplastics enter lung cells and cause inflammation remains unknown. In this project, we study how nanoplastics enter lung cells through A549 cells (lung cells).

1) Grow A549 cells and manage cells.
2) Study how nanoplastics enter A549 cells using endocytosis assays.

Alzheimer's Disease (AD) is a neurodegenerative disorder affecting the global elderly population and currently lacks effective treatment. APOE gene is the strongest genetic risk factor for AD, whereas the APOE4 allele increases disease risk, APOE2 has a protective effect. Recent large-scale genetic analyses have identified several protective APOE variants that significantly reduce AD risk, though the mechanisms of this protection remain unclear. This project aims to employ in vitro biochemical assays and AAV-mediated gene delivery in AD mouse models to investigate whether and how these protective variants modify apoE properties and alleviate AD-related pathology, which could deepen our understanding of the mechanisms behind APOE protective variants in AD.

1) Review literature to understand the project and find knowledge gaps.
2) Assist in experimental design and protocol setup.
3) Conduct lab work, including data collection and analysis.
4) Collaborate on troubleshooting and result discussions.
5) Create reports and presentations on findings.

Preparation of high-quality cryo-specimen is critical to achieve high resolution in single particle cryo-EM studies. This project aims to optimize cryo-specimen preparation by introducing chemical materials.

1) Explore new approaches to optimize traditional plunge freezing.
2) Evaluate different methods by using Cryo-ET and other EM techniques.

Peroxisomes are organelles for the processing of specific fatty acids. Genetic mutations that disrupt peroxisomal function cause varying degrees of liver dysfunction, developmental delay, and neurological defects that impair one's ability to move, see and hear. In this project, mutations that mimic those found in human patients will be introduced into C. elegans. The transgenic animals could potentially be used for genetic and drug screens to identify therapeutic strategies for treating peroxisomal disorders.

1) Design and generate DNA constructs for making transgenic animals.
2) Design and generate DNA constructs for RNAi.
3) Image transgenic or RNAi treated animals by confocal microscopy.
4) Summarize results in an oral presentation.

Prof Yuanliang ZHAI

All living organisms, from single-celled yeast to complex human beings, rely on cell division to propagate. During each cell division, accurate replication of the DNA genome is crucial to ensure that each daughter cell inherits an identical set of genetic information. Any defects in or mis-regulation of this process can result in genetic diseases, developmental abnormalities, and cancers. Therefore, elucidating the molecular basis of DNA replication is essential for gaining critical insights into the pathogenesis of relevant diseases and identifying potential targets for their effective therapeutics. This project aims to utilize structural and functional approaches to understand the mechanisms regulating DNA replication in a context of chromatin.

1) Lead one of the research components of the project, gather and review existing literature, and design and implement experiments.
2) Analyze and interpret the data collected during the experiments.
3) Document the project's progress and findings, write the report, and prepare presentations to share the findings.

Alzheimer’s Disease (AD) is defined by the presence of extracellular plaques from amyloid-beta aggregation and intracellular neurofibrillary tangles that are composed of abnormal tau aggregation. Interestingly, several apolipoprotein E (apoE) receptors can mediate the uptake and spread of amyloid-beta and tau. In this project, we aim to understand the mechanisms by which apoE receptors in various brain cell types impact the seeding and propagation of amyloid-beta and tau.

1) Review literature to understand the project and identify knowledge gaps.
2) Support experimental design and protocol development.
3) Perform experiments, including data collection and analysis.
4) Collaborate on troubleshooting and discuss results.
5) Prepare reports and presentations to communicate findings.

The animals form memory and forget memory to adapt themselves in the changeable environment. By employing proteomics in Drosophila melanogaster, we uncovered the synaptic proteome significant for memory. The project will aim to visualize memory trace by focusing on the synaptic proteins that dynamically change according to the animals’ experience.

1) Conduct optogenetic stimulation.
2) Culture the flies.
3) Dissect and image flies.

This project aims to develop deep learning algorithms and image processing techniques for medical image analysis, including various medical image modalities such as computed tomography (CT), x-ray, magnetic resonance imaging (MRI), whole-slide pathology, fundus, Optical Coherence Tomography (OCT) and Optical Coherence Tomography Angiography (OCTA), etc. Students are also encouraged to propose their new ideas within the scope of this topic. Topics of mutual interests can be flexible. High-quality research papers are expected if possible.

1) Develop algorithms and draft papers.

Diffusion model with deep learning has achieved significant progress in image generation and analysis. In this project, students will investigate state-of-the-art diffusion methods for medical imaging and analysis, such as super-resolution, semantic recognition, etc.

1) Develop advanced diffusion models and publish top-tier papers.

The transfer of high-quality hexagonal boron nitride (hBN) with minimal surface contamination is essential for advanced characterization techniques, such as Scanning Tunneling Microscopy (STM), which demand ultra-clean surfaces. In this study, a transfer method will be used for chemical vapor deposition (CVD) hBN using Cellulose Acetate Butyrate (CAB) as a supporting layer, optimized to maintain the cleanliness and integrity of the hBN surface post-transfer. CVD-grown hBN on polycrystalline copper (Cu) foil will be transferred onto target substrates with CAB, to produce uniform, continuous monolayer and few-layer films with low defect density.

1) Master transfer processes, ensure proper handling of materials, and adhere to procedural protocols.
2) Analyze data from experiments.

Medical data sets are difficult to obtain due to the complicated nature of the healthcare system. For example, different hospitals may only be able to access the clinical records of their own patient populations. An intuitive idea is to learn those data without sharing. Federated learning (FL) is a learning paradigm to address the problem of data governance and privacy by learning models collaboratively without exchanging the data. This project will explore the federated learning in healthcare, e.g., medical image analysis. Several topics will be investigated under FL setting, such as domain generalization, differential privacy, and personalization.

1) Develop advanced algorithms and draft papers.

The development of efficient catalysts for oxygen reduction reactions (ORR) is critical for advancing the performance of lithium-oxygen (Li-O2) batteries. In this project, we will investigate graphene-supported single-atom catalysts designed to enhance the ORR activity and selectivity in Li-O2 systems. The electrochemical performance of the catalyst will be investigated after its synthesis.

1) Investigate the synthesis of graphene-supported single-atom catalysts.
2) Analyze data from experiments.

In clinical practice, heterogenous multi-modality data (e.g., radiology images, clinical history, reports, etc.) will be collected for cancer diagnosis and prognosis. With the development of advanced computational methods, various biomarker features can be extracted and mined for cancer analysis. This project aims to develop state-of-the-art deep learning methods for cancer diagnosis, treatment outcome prediction and prognosis by mining multi-modality data. Students are encouraged to publish high-impact research papers.

1) Develop advanced algorithms and draft papers.

Large pre-trained models, trained on vast and diverse datasets, have recently demonstrated significant advancements in vision and language domains. In computational pathology, these models hold transformative potential to improve diagnostic accuracy, reduce the manual workload on pathologists, and optimize clinical workflows. However, adapting these models to specific downstream pathological tasks demands targeted training to address the unique complexities of each task, often presenting computational and performance challenges. This project aims to leverage off-the-shelf pre-trained vision and language models, fine-tuning them to optimize their performance in downstream tasks. By employing multi-modal prompt tuning techniques, this project seeks to adapt these models both efficiently and effectively for a variety of downstream pathological tasks.

1) Curate and preprocess diverse datasets to develop multi-modal fine-tuning datasets.
2) Create and optimize multi-modal prompts to facilitate efficient interaction between vision and language components, tailoring them to specific clinical tasks.
3) Design and implement a pipeline for fine-tuning pre-trained models.
4) Conduct experiments across various downstream clinical tasks.
5) Analyze experimental results and write the paper.

Controllable synthesizing of large-scale van der Waals metallic materials and tuning their atomic and electronic structures is nowadays of particular interest due to the potential applications for electrodes with low contact resistance. In this study, we will introduce a phase-controlled growth method to synthesize large scale van der Waals metallic Mo6Te6 films by annealing the large scale 2H-MoTe2 films with high quality. The annealing temperature and mixture of Ar and H2 are critical factors to influence the transformation from 2H-MoTe2 to Mo6Te6 films. Additionally, Raman spectra will be utilized to characterize the types of synthesized materials to identify the successful transformation.

1) Master the chemical vapor deposition (CVD) process, ensure proper handling of materials, and adhere to procedural protocols.
2) Perform characterization techniques, including optical microscopy and Raman spectroscopy.
3) Analyze data from experiments.

Medical image analysis plays a crucial role in healthcare by providing a non-invasive way to diagnose and treat various medical conditions. With the advent of deep learning techniques, computer vision-based medical image analysis has achieved remarkable success in recent years. However, the current state-of-the-art methods are limited by the availability of large annotated datasets and the need for expert domain knowledge. This research proposal aims to explore the application of vision language models and large language models (e.g., CLIP, ChatGPT) in medical image analysis to address these limitations and improve the accuracy and efficiency of diagnosis.

The proposed research can have numerous applications in the medical field. It can aid in the automated detection of various diseases such as cancer, heart disease, and neurological disorders by analyzing medical images such as X-rays, MRIs, and CT scans. It can also assist doctors and radiologists in making more accurate and timely diagnoses, leading to better patient outcomes. The research can also be extended to drug discovery and personalized medicine by analyzing molecular images.

Self-attention has gained remarkable success in sequence-based recognition tasks. In this project, we will explore the power of self-attention based deep learning, i.e., transformers, in medical imaging and analysis, e.g., super-resolution, restoration, classification, segmentation and registration, etc.

1) Develop algorithms and draft papers.

Pervasive deployment of sensors has enabled the tracking of moving objects, giving rise to real-time trajectory streams of huge volume.  This phenomenon creates opportunities for online detections of clusters and anomalies for applications such as traffic management, pedestrian flow, marine transportation, and animal migration.  A challenge is thus to compress the trajectory streams in an online, efficient, and effective manner so that clusters and anomalies can be found quickly.  This project is about the implementation of streaming simplification algorithms for Fréchet distance and discrete Fréchet distance.

1) Study the streaming simplification algorithms developed by the PI and collaborators.
2) Implement the algorithms.
3) Experiment.

In the last decade, deep learning has achieved dramatic success in pattern recognition tasks. However, the characteristic of data hungry training can pose huge challenges in real-world tasks. In addition, the interpretation and domain generalization capability can be vulnerable in existing methods. In this project, we aim to explore to tackle these challenges by developing data-efficient training, interpretable and domain generalizable deep learning techniques. Publish high-quality research papers if possible.

1) Propose new ideas within the scope of this topic; topics of mutual interests can be flexible.
2) Develop algorithms and draft papers.

This project aims to develop and optimize techniques for efficient high-dimensional search within vector databases. As data continues to grow in dimensionality, traditional search methods often become inefficient and slow. The project will explore advanced indexing methods, dimensionality reduction techniques, and approximate nearest neighbor search algorithms. Students will work on implementing these techniques to enhance the speed and accuracy of similarity searches in applications such as recommendation systems, image retrieval, and natural language processing.

1) Research Engineer: Investigate and implement various algorithms for high-dimensional search, explore existing literature, and develop prototypes.
2) Data Analyst: Evaluate the performance of implemented algorithms, analyze search efficiency, and conduct experiments to compare different approaches.

In this project, students are guided to conduct research in the database field. Depending on students’ strengths, they are given particular database problems. For example, if they are good at geometry, some candidate problems include “how to find a nearest restaurant from a given location” and “how to find a shortest path from a source to a destination.” If they are good at programming, one candidate problem is “how to return results efficiently when users issue some queries.”

Requirement: GPA/CGA at least 3.7 (out of 4.0).

1) Study some important problems in the field.
2) Implement some important algorithms in the field.
3) Conduct research.

In this project, students are guided to conduct research in the data mining field (or the knowledge discovery field). Depending on students’ strength, they are given some particular data mining problems. For example, if they are good at optimization, some candidate problems include “how to find the best discount rate for a shop in order to attract more customers” and “how to promote products for a shop”. If they are good at graph, one of the candidate problems is “how to find similar friends in Facebook”.

Requirement: GPA/CGA at least 3.7 (out of 4.0).

1) Study some important problems in the field.
2) Implement some important algorithms in the field.
3) Conduct research.

This project focuses on enhancing the capabilities of language models by integrating a retrieval mechanism that utilizes a vector database. The goal is to improve the accuracy and relevance of generated responses by allowing the model to fetch pertinent information from a vast dataset. By combining retrieval techniques with generative models, students will explore how to optimize the architecture for better performance in specific applications, such as question answering and conversational agents.

1) Data Engineer: Prepare and manage datasets, including data cleaning and preprocessing.
2) Machine Learning Engineer: Implement and fine-tune the retrieval-augmented generation model, exploring different architectures and algorithms.

Emerging non-volatile memory, sensors, energy harvesters, and data processing accelerators based on emerging materials are promising for AI and IoT applications. To design functional circuits, compact models need to be established for these emerging devices. The objective of this project is to develop a compact model for emerging devices such as magnetic random-access memory to build circuits for fully investigating the potential applications based on these emerging devices and systems.

1) Read literature independently and work with the supervisor and PG students to build compact models and design circuits.

Quantum computing has demonstrated advantages in specialized tasks over classical supercomputers. In the future, quantum computing is envisioned to provide essential computational support for medicine design, encryption, secure communication, quantum machine learning, and beyond. At this moment, superconducting qubit devices are the upfront contender for future quantum integrated circuits. To scale up the system, electronic design automation (EDA) tools are essential to reduce the design cycle time and improve yield. In this project, we will explore IBM Qiskit Metal, a first-of-its-kind and open-source project for quantum computing EDA, and investigate how to build more powerful EDA tools. The objective of this project is to understand the state of the art of EDA for superconducting quantum computing and design a prototype integrated circuit for basic single-qubit and two-qubit gate operations.

1) Understand the basics of superconducting circuits, qubits, and quantum computing.
2) Install IBM Qiskit Metal and run tutorial examples.
3) Design qubits and reproduce qubit characteristics.
4) Design qubit gates.
5) Design hybrid quantum systems including photon-magnon, photon-qubit, phonon-magnon, etc.
6) Publish research papers.

Combinatorial optimization problems are central to numerous scientific and industrial fields but are often computationally intensive and challenging to solve. The project, "Efficient Combinatorial Optimization Solver Based on Novel Spintronic Devices", aims to address these challenges by utilizing the unique properties of spintronics. Spintronics offers a paradigm shift in computation, leveraging electron spin rather than just charge, enabling devices that are faster, more energy-efficient, and capable of parallel processing. Our research focuses on developing spintronic devices that can inherently process combinatorial optimization problems, potentially outperforming traditional computing methods. This research project combines physics, engineering, and computer science to create a novel optimization solver that is fast, reprogrammable, scalable and energy-efficient. The selected UROP students will be instrumental in pushing the boundaries of what is possible with spintronic technology, helping to solve complex optimization problems that are currently beyond the reach of conventional computational methods.

1) Search and review the related literature and work with the supervisor and PG students.
2) Algorithm Implementation: Write and optimize code to implement and test combinatorial optimization algorithms relevant to the project's goals.
3) Hardware System Design and Construction: Collaborate in the design and assembly of a hardware system that emulates the spintronic optimization solver.
4) Experimental Testing: Perform hands-on experiments to test and characterize the performance of spintronic devices, documenting and analyzing results to guide further development.

The mainstream computer system has separate memory and logic units, which is the so-called von Neumann architecture. Data-centric computing such as artificial intelligence (AI) algorithms demands constant data transfer between logic and memory units, resulting in a huge delay and energy cost. The integration of computing circuits and emerging non-volatile memory or so-called in-memory computing is promising for accelerating AI algorithms. Test equipment is required for characterizing functional nonvolatile memory devices and systems. The objective of this project is to develop a prototype automated test equipment for intelligently investigating nonvolatile memory and sensors.

1) Understand basic components of the given PCB system, including power management circuits, DAC, ADC, STM32, VCCS, communication module, etc.
2) Understand the basics of non-volatile memory technology and its importance for artificial intelligence hardware.
3) Reproduce the PCB system using the schematic of the given PCB system, test it and finish documentation.
4) Design a PCB system with better performance in terms of speed, power, and parallelism.
5) Fabricate the designed PCB system, test it and finish documentation.

We use a Michelson interferometer setup to achieve the Fourier-domain OCT. The student will assist in the development and implementation of algorithms for processing and analyzing the OCT data.

1) Collaborate with team members on the experimental design and data collection.
2) Conduct literature research.
3) Develop and optimize image processing algorithms for the OCT data.

We will develop an optical reservoir computing system using our fabricated heterogeneous integrated III-V-on-Si lasers to demonstrate a 2-bit XOR operation. The student will take part in the investigation and construction of the computing system and help develop the data processing code.

1) Conduct a literature review of reservoir computing.
2) Participate in the optical measurements of the lasers.
3) Participate in constructing of an optical reservoir computing system.
4) Develop the data processing code for a 2-bit XOR operation.

This project investigates quantum interference using the Hong-Ou-Mandel (HOM) experiment, where photon pairs from a periodically poled lithium niobate (PPLN) waveguide are used as quantum light sources. We will analyze quantum interference in a free-space optical setup for applications in quantum metrology and quantum computing.

1) Collaborate with team members on experimental design and data collection.
2) Conduct literature research.

The emerging in-memory computing technologies holds the great promise for achieving ultra-low power consumption in future efficient architectures. Among that various implemented hardware, the novel spintronic devices have emerged as an excellent candidate due to their rich nonlinear dynamics and non-volatile memory effect. The proposed project, ‘spintronic dynamics for reservoir computing’, aims to develop a spintronic-based computing diagram and enhance its computing capacity by controlling the spintronic dynamics. The nonlinearity and complexity dependency of spintronic for information processing capacity have been frequently observed, such as the edge of chaos, but still lacking the underlying principles behind those phenomena. Our research mainly focuses on computing optimization for spintronics-based systems. The selected UROP student will have the opportunity to deeply learn the nonlinear dynamics for spintronics and explore methods to enhance their computing capabilities, contributing to future spintronic-based in-memory computing technologies.

1) Review the related literature and work with the supervisor and PG students.
2) Modelling: Utilize coding languages such as Python, MATLAB, or Mumax3 to simulate spintronic dynamics using the Landau-Lifshitz-Gilbert (LLG) equations. Build an understanding of the nonlinear dynamics in input-driven spintronic models.
3) Computing: Develop and implement a reservoir computing algorithm implemented by the spintronic model. The benchmarks will be utilized to evaluate the computing capacity of systems with different input encoding. The physical relationship between input setting and entire information processing capacity is encouraged to be developed more.
4) Performance: Controlling the spintronics dynamics, guided by relevant theories, to obtain the better performance, such as higher accuracy or robustness. One interesting potential target is achieving the prediction of spintronic chaotic dynamics by utilizing the order spintronic-based system.

VR/AR applications introduce a completely different traffic pattern to the transport protocols – it sends bursty, unpredictable traffic, it requires consistently low latency, and it also asks for as many throughput as possible. Existing congestion control, video codec, and network protocols are not designed for such a scenario, which motivates us to redesign the network architecture to better support the ultra-low latency applications such as VR/AR.

1) Brainstorm with the PIs and investigate how to improve the performance of VR/AR applications.
2) Set up the testbed for ultra-low-latency video streaming.

This research project aims to explore the feasibility and efficiency of 3D printing using lunar soil (regolith) to create sustainable structures for future lunar habitats. The study will focus on analyzing the properties of lunar regolith to determine its suitability for 3D printing and developing optimized printing techniques that can withstand the harsh lunar environment. Additionally, the project will assess the environmental impact and resource efficiency of utilizing in-situ materials compared to traditional construction methods.

1) Engage in various roles throughout the project, including conducting material analyses, developing printing methodologies, and creating prototypes of lunar structures.
2) Collaborate with a multidisciplinary team of researchers, engineers, and space scientists to ensure a comprehensive approach to the challenges posed by lunar construction.
3) Document findings, present results, and contribute to the overall project strategy.

This project aims to develop an additive manufacturing printing nozzle and feeding system of continuous carbon fiber. The printing nozzle is expected to take in both thermoplastics and continuous fiber stocks with tunable mix ratios. The printing nozzle and feeding system will be attached to robotic arms and motion stages to perform a functional 3D print of fiber-reinforced parts. The students will also work with the Ph.D. students to develop trajectory planning and control software.

1) Design the printing nozzle with multiple stock feeding material, continuous fiber cutter, and other relevant mechatronics devices.
2) Control the feeding system and tune various process parameters.
3) Incorporate this printing solution into an existing machine tool feed drive or robotic arm.

This project aims to develop an additive manufacturing printing nozzle and feeding system of metal in the form of wire stock. The wired metal is expected to be melted by electromagnetic induction heating at the nozzle, followed by a conventional deposition process. The printhead will be incorporated into an existing machine tool feed drive to enable additive manufacturing functionality. The project may also lead to a hybrid manufacturing process (combination of additive manufacturing and machining) in collaboration with the faculty members and other Ph.D. student.

1) Design the printing nozzle with the electromagnetic heating and wire feeding system.
2) Optimize the printing parameters for efficiency and speed.
3) Incorporate this printing solution into an existing machine tool feed drive.

The project aims to develop software and hardware augment reality (AR) solution to gamify regular life and sports training, particularly in tennis training/match. The students are going to develop offline and real-time tennis training video analysis software to quantify the accuracy, speed, and spin of the hit, with additional visual illustration, evaluation, and player motivation system.

1) Generate the video data for the software development.
2) Use existing video analysis packages to determine the tennis ball trajectory, speed, and spin.
3) Design and formulate the evaluation system for the hit quality.
4) Program the augment reality video visual effect to highlight the evaluation.
5) Program the real-time hardware to further enhance the training experience.

Dragonflies can remain stationary in one position in the air and change any flight direction instantaneously. This outstanding maneuverability primarily arises from its unique wing geometry, aeroelastic property, and control strategies. For the study of dragonfly wings' properties, reliable fabrication techniques that emulates the stiffness variations at different locations of the wing are needed. This project aims to achieve this goal of dragonfly wing fabrication via continuous carbon fiber 3D printing technologies. The applicants are expected to choose appropriate materials and manufacturing technologies to fabricate the dragonfly wings and perform preliminary tests on their mechanical properties.

1) Operate and program the continuous carbon fiber printer Anisoprint Composer A4.
2) Collaborate with other Ph.D. students to finalize the design and print pattern.
3) Explore different materials (PLA, ABS, Nylon) and manufacturing technologies.
4) Fabricate the dragonfly wings.
5) Perform preliminary mechanical testing.

The project focuses on applying Generative AI in design and manufacturing processes to automate design generation, optimize designs, and foster creativity. It involves developing AI algorithms and training them on relevant datasets to generate innovative and efficient designs that meet specific criteria. The project aims to accelerate the design process, improve manufacturing efficiency, and unlock new possibilities for optimized product designs.

1) Conduct research on Generative AI in the manufacturing domain.
2) Prepare datasets for training AI models.
3) Develop or adapt AI algorithms for generative design.
4) Train and evaluate AI models for generating designs.
5) Experiment and iterate on the AI models for improved performance.

The project aims to develop a simulation framework for the 3D printing process. The simulation will involve creating a virtual model that replicates the additive manufacturing process, taking into account various parameters and factors that affect the final printed object's quality and performance. The simulation will consider factors such as material properties, printer configuration, layer deposition, cooling rates, and structural constraints. By accurately modelling these aspects, the simulation will provide insights into the behavior of the material, heat transfer, and the interaction between the printed layers. The project will focus on predicting and analyzing potential issues that may arise during the printing process, such as warping, distortion, residual stresses, and the need for support structures. By simulating these scenarios, the project aims to optimize the printing parameters, modify the design, or suggest alternative materials to improve the quality and efficiency of the 3D printing process.

1) Conduct research on the existing simulation methods and techniques used in 3D printing.
2) Develop or customize software for the simulation framework.
3) Create virtual models of the 3D printers, materials, and objects to be printed.
4) Analyze and experiment with different parameters that affect the 3D printing process, such as material properties, printer settings, and cooling rates.
5) Evaluate the simulation results and compare them with real-world 3D printing outcomes.

Sheet metal incremental forming is a technique where the sheet is formed into the final piece by a series of small incremental deformations. It eliminates the die in the forming process, enabling rapid prototyping and low-volume manufacturing with low cost. The student will have the opportunity to use FEA software (Abaqus/Ansys) to simulate the process and develop a reduced-order model describing the process, further contributing to the trajectory generation and control of the incremental forming process.

1) Use simulation software (Abaqus/Ansys) to model the deformation in the single- or multiple-point sheet metal forming process.
2) Exploit knowledge of continuous mechanics and structural analysis to formulate a reduced-order model.
3) Develop advanced trajectory generation methods of the incremental forming process with enhanced accuracy and efficiency under the reduced-order model.

In this project, we examine whether there are personal incentives of mutual fund managers when they make ESG decisions. We use data from US mutual fund managers.

1) Help with cleaning and processing data.
2) Go through matches generated using AI and machine learning methods.

Environmental, Social, and Governance (ESG) has become a very important topic around the globe due to climate change as well as a rise in social and environmental awareness. However, it is still not clear whether the most important players in financial markets, the asset management companies (e.g., hedge funds), care about ESG. This project aims to understand whether certain types of asset management companies have specific preferences to ESG. This is a joint project with professors from the University of Chicago, University of California at Berkeley, and Boston University.

1) Collect data, manually match datasets, and read various reports related to ESG from, for instance, the U.S. Securities and Exchange Commission.
2) Engage in other data-related tasks.

Environmental, Social, and Governance (ESG) has become a trending topic. Although much research has been conducted recently, the role of key financial players, such as institutional money management companies (e.g., pension funds, hedge funds) in ESG outcomes is not fully clear. By building on findings from a joint project with professors from the University of Chicago, University of California at Berkeley, and Boston University, this project will allow students to use institutional ownership data, such as portfolio holdings, shareholder voting data, PAC data, and other related datasets.

1) Conduct a literature review.
2) Help with data cleaning and matching.

This project examines the roles of heuristics and biases in judgment and decision-making. We will conduct experiments informed by psychology and economics to examine how people make choices relevant to a business context. As there are multiple ongoing streams of research, the specific topic can be matched with your interests. See David's research to get a sense of the kind of projects you would be involved in.

1) Implement experimental surveys in Qualtrics.
2) Conduct basic data processing and visualization.
3) Classify participants’ responses according to a schema.
4) Participate in research and experimental designs.

There are two competing approaches to scientific progress in the philosophical literature. The epistemic approach defines the nature of scientific progress as the accumulation of scientific knowledge, while the new functional approach construes it as the increase of usefulness of exemplary practices in science. Recently the functional approach has been fruitfully applied to explain cases of progress in astronomy, biology, psychology, and interdisciplinary research. A difficulty case is from the history of chemistry, to which the epistemic approach is often applied. This project aims to examine the application of the new functional approach to the historical development of the periodic table. Such a case study will shed light on the recent philosophical debate on scientific progress and deepen our understanding of the development of 19th century chemistry.

1) Read original scientific texts and secondary (philosophical and historical) literature and develop coherent arguments concerning scientific progress and the development of the periodic table.
2) Write a cohesive philosophical essay on the topic.

Recent philosophical discussions on scientific understanding and scientific progress focus on the context of the natural sciences. The role and nature of scientific understanding and its relation to scientific progress in the social sciences are insufficiently examined. This project aims to examine the role and nature of scientific understanding in the social sciences and shed light on the debate over scientific progress and scientific understanding.

1) Read important and recent work on scientific understanding and scientific progress.
2) Participate in the regular philosophy of science research group and actively take part in the discussion.
3) Assist in organizing the ‘Explanation, Narrative, and Understanding in the Social Sciences’ conference (25 – 27 June 2025).
4) Write a 3,500-word paper on the topic of the project.

This project aims to find solutions to protect the earth through understanding how children learn about the earth. This project offers the first step to identify the cognitive factors that motivate the pro- and anti-environmental behavior observed in children and adults. We will answer questions including but not limited to 1) How do children and adults understand the origins of their food? 2) What's the role of the parents in raising an 'environmentalist"? etc.

1) Attend regular lab meetings.
2) Conduct literature reviews.
3) Prepare experimental material.
4) Participate in recruitment and coordination.
5) Conduct hands-on experiments, both in-person and online.
6) Perform data analyses.
7) Prepare poster and presentation at conferences.
8) Basic data analysis skills (Excel, R, SPSS) are expected. Programming knowledge (e.g., Python, PsychoPy, JSON) is recommended but not required.

Arithmetic ability is one of the key predictors of academic achievement. This study aims to examine the early development of arithmetic skills in pre-primary and primary school children. We focus on understanding 1) the developmental trajectory of different arithmetic skills; 2) factors and predictors that may contribute to and account for the development of arithmetic abilities; 3) how individual differences in cognitive functions may relate and predict future arithmetic performance; 4) what early mathematical education can do to support children's STEM learning.

1) Attend regular lab meetings.
2) Conduct literature reviews.
3) Prepare experimental stimuli and participate in recruitment.
4) Conduct experimental studies on children and adults.
5) Perform statistical analyses and generate research outcomes for conferences.

Whether, what, and how one thinks about the future make substantial consequences on one's actions and anticipation. In early development, how children think about the future determines what they decide to do at the moment. The current project aims to examine the underlying mechanism of future-oriented thinking and the role of future-oriented thinking in a variety of cognitive activities including memory-guided planning, environmental protection and injury-prevention.

1) Attend regular lab meetings.
2) Conduct literature reviews.
3) Prepare experimental material.
4) Participate in recruitment and coordination.
5) Conduct hands-on experiments, both in-person and online.
6) Perform data analyses.
7) Prepare poster and presentation at conferences.
8) Have some basic data analysis skills (Excel, R, SPSS). Programming knowledge (e.g., Python, PsychoPy, JSON) is recommended but not required.

This project uses the increasing availability of digitized public records to examine the long-term social, political, and economic outcomes of dramatic events. Though a variety of public records are increasingly available in digital format, linking individuals across different record sources presents complex methodological challenges. Some issues that we will investigate include the intergenerational impacts of political violence, the effects of mass shootings on the social and political attitudes of the victims, and the behavioural effects of racial segregation and desegregation on economic well-being. Given that there are several streams of ongoing research, the specific topic can be matched to the interests of the applicant.

1) Conduct literature reviews.
2) Collect and clean unformatted quantitative data from the web.
3) Develop machine learning models for automated record matching.
4) Conduct statistical analysis.
5) Assist with the development of new research areas.

This project examines the social contagion of mental health symptoms in the age of social media and places it in historical context. Mass psychogenic illness, also known as mass hysteria, involves the spread of illness symptoms in a population through social means rather than through an infectious agent. Contemporary examples of rapid increases in mental health symptoms among social media communities that have been characterized by experts as mass psychogenic illness are (1) Tourette Syndrome, (2) dissociative identity (multiple personality) disorder, (3) eating disorders, and (4) rapid onset gender dysphoria. We will investigate (1) whether such cases do indeed qualify as mass psychogenic illnesses and, if so, (2) the similarities and differences between modern mass psychogenic illnesses and those of a previous era prior to social media. Given that there are several streams of ongoing research, the specific topic can be matched to the interests of the applicant.

1) Conduct literature reviews.
2) Collect and clean unformatted quantitative data from the web.
3) Develop machine learning models for automated record matching.
4) Conduct statistical analysis.
5) Assist with the development of new research areas.

Working memory updating allows maintaining an up-to-date representation of the world. How is information updated in working memory, and what factors influence the updating performance in children? This project aims to understand the early development of information updating in working memory.

1) Participate in lab meetings.
2) Conduct literature reviews.
3) Prepare experimental stimuli.
4) Conduct experimental studies on children and adults.
5) Perform statistical analyses and generate research outcomes for conferences.

This project examines the effects of structural factors and natural experiments on political phenomena. For political phenomena ranging from civil wars to democratic elections, we will investigate the ways in which public policies and economic conditions causally impact political outcomes. Given that there are several streams of ongoing research, the specific topic can be matched to the interests of the applicant.

1) Conduct literature reviews.
2) Collect and clean unformatted quantitative data from the web.
3) Conduct statistical analysis.
4) Assist with the development of new research areas.

This research project aims to explore and enhance the user experience (UX) and user interface (UI) of Articulator, a new digital platform designed for AI-assisted art creation. This platform has applied for TSSSU funding. The focus of the internship will be on understanding how users - both professional and emerging artists - interact with generative AI tools and the overall platform. Undergraduate students will take on various roles, including UX researchers, UI usability testers, and data analysts. They will conduct user interviews and surveys to gather insights, analyze existing designs, and develop wireframes and prototypes. Usability testing will be performed to evaluate these prototypes, collecting feedback to refine the user experience further. The project will also integrate HKUST licensed inventions in embedding visual 2-D barcode technology and decentralized identification (DID) to enhance authenticity and traceability of artworks.

1) Conduct user interviews and surveys to gather insights on user needs, analyze existing designs, and create user personas and journey maps.
2) Conduct usability testing sessions to evaluate UI prototypes, collect user feedback, and analyze results to provide recommendations for improvements.
3) Develop wireframes and prototypes based on research findings, ensuring designs enhance usability and align with the platform's brand identity.
4) Analyze feedback and testing data to identify trends, prepare reports summarizing findings, and track key performance indicators related to user engagement.

The study would involve analysing financial data and performance metrics of companies or investment portfolios that prioritize environmental, social, and governance (ESG) factors in their investment decisions. This is a joint project with researchers from Harvard University.

1) Help organize large datasets.
2) Run preliminary statistical analyses on the cleaned-up data.

Multi-axis 3D-printing offers several advantages over conventional 3D-printing, such as support-free printing, enhanced surface finishing, and control of anisotropic properties for strength reinforcement. Recently, The centre for perceptual and interactive intelligence (CPII) has developed a first multi-axis 3D-printer that uses Digital Light Processing (DLP). This technology opens up new opportunities, such as the ability to print highly flexible materials. In this project, you will explore applications of the described technology in the field of soft robotics. This project will explore the applications of multi-axis DLP in the field of soft robotics.

1) Work closely with experts in multi-axis 3D-printing and soft robotics.
2) Come up with new soft robot designs with unique capabilities that could previously not be manufactured due to manufacturing constraints.

The PI and his team have previously designed a soft hydraulic gripper inspired by the seahorse tail for underwater manipulation. In this project, the UG students will explore the effect of key design parameters on the performance of the soft hydraulic gripper. The students will deploy additive manufacturing to fabricate a range of seahorse-inspired soft grippers with varying design parameters. Then, the students will build an experimental setup to characterize the different designs in terms of holding force. The students will select the most suitable design for deployment on a remotely operated underwater vehicle (ROV) and demonstrate its performance in an underwater trash collection task.
This project is related to the Sustainable Smart Campus project "Sustainable Smart Marine Grid (SSMG) - Solar-powered base stations for continuous underwater surveying and trash cleaning”.

1) Manufacture a range of soft actuators with varying design parameters.
2) Characterize the performance of the actuators in terms of holding force.
3) Mount the actuators on an ROV.
4) Control the ROV and gripper to perform an underwater trash collection task.

OpenFish is an open-source soft robotic fish which is optimized for speed and efficiency. The soft robotic fish uses a combination of an active and a passive tail segment to accurately mimic the thunniform swimming mode. In this project, we aim to improve upon the OpenFish design by:
- Implementing a solution for wireless underwater communication with the soft robotic fish
- Implementing mechanisms for controlling the swimming depth and direction of the soft robotic fish.
These improvements will enable the deployment of the soft robotic fish for underwater exploration.

1) Implement a wireless underwater communication system.
2) Implement mechanisms for depth and direction control.

This project is related to the Smart Sustainable Campus Project “Sustainable Smart Marine Grid (SSMG) - Solar-powered base stations for continuous underwater surveying and trash cleaning”. Within this project, the investigators propose a system to collect underwater trash at the HKUST seafront area. The proposed system is composed of an unmanned surface vehicle (USV) acting as the major travel device of the system. The USV is connected to an unmanned underwater vehicle (UUV) that is equipped with cameras and a gripper to detect and collect underwater trash. This UROP project will focus on the development of the unmanned surface vehicle.

1) Design, build, and test an unmanned surface vehicle (USV) equipped with PV-panels, surface antenna, and underwater GPS base station.

This project will examine the electrification of buses in Hong Kong by estimating the energy demand for all regular bus routes and identifying those with high energy consumption. The project will gather GPS data from Hong Kong bus routes, including real-time velocity and gradient profiles of moving buses at one-second intervals that will also record all starts and stops. This data will be incorporated with geospatial road slope data and educated assumptions regarding passenger loads and seasonal air-conditioning usage to predict the electricity demand for each route using an energy calculation model.

1) Analyze driving profile data collected from GPS devices, including correcting deviated GPS points based on road polygons.
2) Monitor GPS data quality and report to the supervisor.

This study will investigate how variation in institutional control of local forest resources impacts the implementation of mangrove conservation and restoration as a climate change adaptation strategy. Specially, this study will consider whether the management regime (i.e., centralized or decentralized to the community) and the local usage rights regime (i.e., degree of community use or reservation) have empirically observable, causal relationships with outcomes in terms of mangrove cover change, following their prioritization for climate change adaptation in India after the 1999 super cyclone in Odisha.

1) Compile and assess socioeconomic datasets from the India National Data Archive.
2) Compile and analyze mangrove cover data from Global Mangrove Watch.
3) Assist in other econometric data analysis, scientific writing, and academic presentations.
4) Merit co-authorship of scientific publication(s) depending on the level of continued involvement.

Urban heat island (UHI) effects are prevalent in Hong Kong and also disproportionately impact disadvantaged socioeconomic groups. Parks, especially green spaces, have been found to mitigate UHI effects in Singapore, Beijing, Seoul, and many other cities. Temperature and UHI effects will intensify in the future due to climate change. Using an existing survey dataset on socioeconomic characteristics and open space visitation, complemented with additional geospatial UHI and climate data, this study seeks to answer the following research questions: 1) Can we confirm previous results that UHI disproportionately affects disadvantaged groups (identified by low income, single adults, occupation)? 2) How does differential thermal impact from UHI affect demand for public open spaces? (while controlling for demographic, socioeconomic, and geospatial characteristics) 3) How will expected temperature increases from climate change impact public open space demand?

Additionally and separately, by analyzing satellite remote sensing data, this study will assess the UHI mitigation effect of different types of vegetation coverage in Hong Kong (e.g., lowland forest, montane forest, deciduous plantation, shrubland, shrubby grassland, and combinations thereof) within undeveloped areas such as country parks in order to evaluate their differential UHI mitigation potential within and beyond their borders.

1) Collect, collate, and analyze publicly available temperature data (USGS Landsat Surface Temperature products) to locate UHI hotspots, calculate UHI intensities, and/or correlate the relationship between UHI mitigation and vegetation cover.
2) Assist in other econometric data analysis, scientific writing, and academic presentations.
3) Merit co-authorship of scientific publication(s) depending on the level of continued involvement.