- LLM Unlearning: Proposed Smoothed Gradient Ascent (SGA) for more stable and effective unlearning in LLMs
- OFFSIDE: Presented a benchmark for multimodal unlearning based on football transfer rumors
- GUARD: Introduced a generation-time unlearning framework for LLMs
- REVEAL: Developed a framework using vision-language models to identify and fix missing and noisy labels in image classification benchmarks
Research Experience
- Research Assistant at HKUST-GZ, supervised by Jiaheng Wei and Zixin Zhong
- Researcher at Irisk Lab, UIUC, leading a project on using machine learning models to help businesses assess risks
Education
- 2024.08 - Present: Master, University of Illinois Urbana-Champaign (UIUC), IL, USA
- 2019.09 - 2023.06: Undergraduate, Central University of Finance and Economics, Beijing, China
Background
- Research Interests: Noisy Labels, LLM Unlearning, Multi-Agent Reinforcement Learning
- Specialization: Data Science and Finance
- Brief Introduction: Currently a research assistant at the Hong Kong University of Science and Technology (Guangzhou), while pursuing a Master's degree at UIUC and interning.
Miscellany
- Passionate about music, served as a piano accompanist and tenor vocalist in the Choir of Central University of Finance and Economics, and composed several musical works