Education
- Bachelor of Arts in Computer Science & Statistics, McGill University, 2020
Dean’s Honours List (Top 5%), CGPA: 3.92/4.00, Major GPA: 3.96/4.00- Selected Math Coursework: Advanved Probability Theory, Optimization, Mathematical Machine Learning, Stochastic Processes, Calculus, Linear Algebra
- Selected Computer Science Coursework: Applied Machine Learning, Natural Language Processing, Computer Vision
Research Positions
- Department of Finance, the Chinese University of Hong Kong, 2021 - Now
- Research Assistant
- Supervisors: Prof. Darwin Choi, Prof. Zhenyu Gao and Prof. Wenxi Jiang
- Quebec AI Institute (MILA) & McGill University, 2019
- Machine Learning Research Intern
- Supervisors: Prof. Joelle Pineau and Prof. Audrey Durand
Research Experiences
- Financial Advisors’ Awareness of Climate Risks: A Textual Analysis Perspective
Co-authors: Prof. Darwin Choi, Prof. Zhenyu Gao and Prof. Wenxi Jiang
Keywords: Climate Finance; Investor Behaviour; Textual Analysis Details
- Preprocess hundreds of gigabytes of PDF documents into TXT format.
- Design a machine learning pipeline for iterative labelling to construct a high-quality training dataset.
- Fine-tune BERT to classify unlabelled corpus and achieve 85+% test accuracy.
- Construct a climate-awareness measure to quantify firms’ realization of climate risks.
- Conduct robust regressions to analyze how climate-awareness affects fund portfolio holdings and return performance.
- A Theoretical Analysis of UCT with Laplace Bound
Co-authors: Prof. Joelle Pineau and Dr. Audrey Durand
Keywords: Machine Learning; Statistical Learning Theory; Optimization Details
- First proved UCT algorithm using the Laplace bound that has a lower regret bound than its counterparts.
- Showed experimental results are consistent with the theoretical regret bound in deep learning settings.
- Bandit Algorithms for Factorial Experiments
Co-authors: Prof. Joelle Pineau and Dr. Audrey Durand
Keywords: Machine Learning; Optimization Details
- Implemented a family of bandit algorithms.
- Investigated various factorial experimental design configurations.
- Concluded that UCT algorithms for factorial experimental designs are robust.
Teaching Experience
Invited Talks
Honours & Awards
- WiML Workshop Travel Grant
- McGill Arts Undergraduate Research Internship Award
- McGill School of Computer Science Research Award
- Tomlinson Undergraduate Award (Teaching Excellence)
- The Rio Tino - Richard Evans International Exchange Award
- Euclid Mathematics Contest School Champion
Technical Skills
- Programming Languages
Stata, Python, R, Matlab, SQL, Shell Script - Machine Learning Packages
PyTorch, Scikit-learn, NLTK, Transformers - Database
Compustat, CRSP, CSMAR, Thomson Reuters