I am a second-year PhD student (2025.8 - now) in the xML Lab at ECE, NUS, advised by Prof. Xinchao Wang. Previously, I received my M.Sc. in Computer Engineering from NUS and my B.E. in Computer Science from Northwestern Polytechnical University.
Research
My research focuses on AI Safety (e.g., data privacy, agent safety) and Multimodal Learning (e.g., vision and language). My recent works explore:
- Agent Safety (ICML'26)
- Data Privacy of Foundation Models (ICML'26, ACL'26 main, NeurIPS'25, Usenix Security'25)
- Efficiency (ICML'26, ICCV'25, ICLR'25, ECCV'24)
News
- [08/2026]: One paper accepted to EMNLP 2026 (Findings): [RPI].
- [08/2026]: Our survey on Diffusion Language Models accepted to TMLR: [dLLM-Survey].
- [07/2026]: Check out our new work on WAM Safety: [BadWAM].
- [07/2026]: One paper accepted to ECCV 2026: [CSS].
- [05/2026]: Three papers accepted to ICML 2026: [Sponge Tool Attack] [dLLM-Attribution] [Mixreasoning].
- [04/2026]: Check out our new survey on VLA Safety: [Arxiv] [GitHub].
- [04/2026]: One paper accepted to ACL 2026 (main): [CoLA].
- [09/2025]: One paper accepted to NeurIPS 2025: [Vid-SME].
- [08/2025]: Starting my Ph.D. at ECE, NUS.
- [07/2025]: One paper accepted to ICCV 2025: [NeuLig].
- [06/2025]: One paper accepted to Usenix Security 2025: [IAM].
Selected Preprints
Selected Publications
Services
Conferences: NeurIPS'24'25'26, ICML'26, ICLR'24'25'26, ICCV'25, ECCV'26, CVPR'26, AAAI'27
Journals: TKDE, ACM Computing Surveys, TMLR, TSNE, etc.
Miscellaneous
When I am not thinking about models, I am fortunate to be supervised by my cat, TunTun (吨吨 in Chinese), who consistently outperforms me in both curiosity and napping efficiency.
