Wei Chen

I am a professor in Institute of Computing Technology, Chinese Academy of Sciences, leading the AI Safety basic research group. I am also adjunct professor in the University of Science and Technology of China, Beijing Jiaotong University, and Nankai University. Before I joined CAS, I was the manager of Computing and Learning Theory Group in Microsoft Research Asia and the co-chair of MSR Asia Theory Center. I obtained my B.S. degree in Statistics from Shandong University in 2006 and my Ph.D. degree in Probability and Mathematical Statistics from Chinese Academy of Sciences in 2011 supervised by Prof. Zhi-Ming Ma. In 2021, I was named by Forbes as one of the 50 Top Women in Tech in China.

My research area is machine learning, especially its basic methods and theory. About ten years ago, I started to investigate how to train large models fast with distributed optimization algorithms, then how to improve the performance of deep learning by understanding its optimization and generalization, and now how to make machine learning safe and trustworthy. My current research interests are causal machine learning and deep learning theory, such as causality-inspried models and o.o.d. prediction, the implicit regularization of the optimization algorithms in deep learning, robust learning and its applications in NLP and CV, etc.


  • Dec 2023: We are organizing ICLR 2024 workshop “Bridging the Gap between Practice and Theory in Deep Learning (BGPT)”. Welcome to join!For more details, please access to the workshop website BGPT.
  • Dec 2023: Our paper “Perturbation-Invariant Adversarial Training for Neural Ranking Models: Improving the Effectiveness-Robustness Trade-Off” is accepted by AAAI 2024.
  • Oct 2023: I am serving as the sponsorship chair of CLeaR 2024. If you are interested to be a sponsor, please contact me.
  • Sep 2023: Our paper “Closing the Gap between the Upper Bound and Lower Bound of Adam’s Iteration Complexity” is accepted by NeurIPS 2023.
  • August 2023:Four of our papers are accepted by CIKM 2023, including one paper on causal-inspired text summaraization and one paper on trustworthy IR.
  • May 2023: Our paper “Convergence of AdaGrad for Non-convex Objectives: Simple Proofs and Relaxed Assumptions” is accepted by COLT 2023.
  • April 2023: We organize the WINDSMATH (Women in Data Science and Mathematics) Seminar. Welcome to join!
  • April 2023:Our paper “Topic-oriented Adversarial Attacks against Black-box Neural Ranking Models” is accepted by SIGIR 2023.
  • Sep 2022: Our paper “Does Momentum Change the Implicit Regularization on Seperable Data?” is accepted by NeurIPS 2022.
  • August 2022: Our paper “Certified Robustness to Word Substitution Ranking Attack for Neural Ranking Models. “ is accepted by CIKM 2022.
  • June 2022: We have two papers accepted by ICLR 2022. One is about policy optimization in RL, the other is about the difussion models.
  • April 2022: we oragnize “Causal Inference and Machine Learning” workshop with AMSS CAS, MSRA Theory Center, and Nankai University, which was held online and in-person at ICT CAS.
  • April 2022: I give a talk at DataSig Lab of Oxford University by the invitation from Prof. Terry Lyons.
  • Feb 2022: I joined Institute of Computing Techonology, Chinese Academy of Sciences.

We Are Hiring!

We are recruiting all-levels (researchers, postdocs and Ph.D. students) who have great passion in basic research to make AI safe. If you have solid CS and/or math background (especially in machine learning), welcome to send your resume to me (chenwei2022 AT ict.ac.cn). Please check this page for details about our group.