Ph.D. in Computer Science
Address: 55 St George St, Toronto, ON, Canada
Email: firstnamelastname[at]cs[dot]toronto[dot]edu
[Google Scholar/ ResearchGate/ Github/ CV]
About me
I am currently a postdoctoral fellow in the Data Science Institute and the Robot Vision & Learning Lab at the University of Toronto, working with Prof. Florian Shkurti. Additionally, I serve as a Faculty Affiliate Researcher at the Vector Institute.
I got my Ph.D. (dissertation with honor) from Laval University in Mar. 2024, where I was supervised by Prof. Mario Marchand. Prior to this, I was a senior algorithm engineer at Baidu and Bytedance. I obtained my master of science in engineering (diplôme d’ingénière) at Institut Polytechnique de Paris - Telecom Paris and my bachelor’s from Chien-Shiung Wu College, Southeast University in China.
Research Goals
I work on making AI more Trustworthy and Efficient. My research focuses on enhancing the generalization of modern deep learning models in out-of-distribution settings and dynamically changing environments, considering potential fairness and bias issues. I see these as fundamental challenges in building responsible AI and advancing toward AGI.
My PhD research tries to interpret basic human intelligence from the perspective of acquiring and exploiting prior knowledge. I believe real intelligence hinges on efficient knowledge modulation—how to effectively learn, store, retrieve, and compose knowledge—integrating high-level reasoning and verification processes.
This perspective drives my broad interests in meta-learning, continual learning, generative models, algorithmic fairness, Bayesian optimization, and reinforcement learning. I explore these areas from both theoretical and algorithmic standpoints, with a strong desire to apply them to high-impact domains such as scientific discovery.
Feel free to email me for potential collaborations or discussions.
News
- [01/25] One paper accepted to ICLR 2025, see you in Singapore.
- [05/24] One paper was accepted to ICML2024.
- [01/24] I passed my thesis defense.
- [09/23] One paper was accepted to NeurIPS2023.