Sizhe Chen’s homepage

Biography

Hi! I am a CS Ph.D. student at UC Berkeley advised fortunately by Prof. David Wagner in Wagner Lab and Berkeley AI Research. I am working concurrently at Meta FAIR with Chuan Guo at San Francisco. I got my M.Eng. and B.Eng. (Summa Cum Laude) from Shanghai Jiao Tong University advised by Xiaolin Huang and also with Cihang Xie.

My research focuses on AI security in real-world applications. I am working on defenses against prompt injections (StruQ, Jatmo), which arise as LLM-integrated applications generally craft LLM inputs by concatenating a trusted prompt from the developer with an untrusted data from the user. Security against prompt injections enables broader use of LLMs, e.g., as an agent. My work on trustworthy vision models is listed in my CV and (previous) SoP.

Outside research, I love to play badminton, lift weights, write blogs, and attend concerts (I directed three 1K-spectator concerts). Feel free to drop me an email to connect!

Selected Publications

• StruQ: Defending Against Prompt Injection with Structured Queries
  Sizhe Chen, Julien Piet, Chawin Sitawarin, David Wagner
  [arXiv’24], [Code]
• One-Pixel Shortcut: On the Learning Preference of Deep Neural Networks
  Shutong Wu*, Sizhe Chen*, Cihang Xie, Xiaolin Huang
  [ICLR’23 (Spotlight)], [Code]
• Adversarial Attack on Attackers: Post-Process to Mitigate Black-Box Score-Based Attacks
  Sizhe Chen, Zhehao Huang, Qinghua Tao, Yingwen Wu, Cihang Xie, Xiaolin Huang
  [NeurIPS’22], [Code]
• Universal Adversarial Attack on Attention and the Resulting Dataset DAmageNet
  Sizhe Chen, Zhengbao He, Chengjin Sun, Jie Yang, Xiaolin Huang
  [TPAMI’22], [Code]
• Subspace Adversarial Training
  Tao Li, Yingwen Wu, Sizhe Chen, Kun Fang, Xiaolin Huang
  [CVPR’22 (Oral)], [Code]

Services

• Conference Reviewer: SaTML’25, CCS’24, ICML’24, NeurIPS’23, ICLR’23/24, CVPR’23/24, ICCV’23, ECCV’22/24
• Journal Reviewer: IEEE TPAMI, Machine Learning, Pattern Recognition
• UC Berkeley CS Faculty Hiring Committee: 2024