Weijie Su 苏炜杰
Professor (on leave)
Wharton Statistics and Data Science Department
and, by courtesy, Departments of Computer and Information Science, Biostatistics, Epidemiology and Informatics, and Mathematics
Co-Director, Penn Research in Machine Learning
University of Pennsylvania
Email: suw AT wharton DOT upenn DOT edu
Member of Technical Staff
OpenAI
Research Interests
(Long-term) Mathematical, compute-light approaches to understanding deep learning and AI
(Current) Statistical foundations of large language models, privacy-preserving machine learning, high-dimensional statistics, mathematical optimization, phenomenological deep learning theory
Education
Ph.D., Stanford University, 2016
B.S., Peking University, 2011
Recent Featured Papers
- Isotropic Curvature Model for Understanding Deep Learning Optimization: Is Gradient Orthogonalization Optimal? W. Su.
- Do Large Language Models (Really) Need Statistical Foundations? W. Su.
- Statistical Impossibility and Possibility of Aligning LLMs with Human Preferences: From Condorcet Paradox to Nash Equilibrium. K. Liu, Q. Long, Z. Shi, W. Su, and J. Xiao.
- The 2020 United States Decennial Census Is More Private Than You (Might) Think. B. Su, W. Su, and C. Wang.
- A Statistical Viewpoint on Differential Privacy: Hypothesis Testing, Representation and Blackwell's Theorem. W. Su.
- A Statistical Framework of Watermarks for Large Language Models: Pivot, Detection Efficiency and Optimal Rules. X. Li, F. Ruan, H. Wang, Q. Long, and W. Su.
- A Law of Next-Token Prediction in Large Language Models. H. He and W. Su.
News
- August 2026. I am deeply honored and privileged to receive the 2026 COPSS Presidents' Award at JSM in Boston.
- November 2025. A paper was invited for presentation at the JRSSB Editors' Invited Session at the 2026 RSS Conference in the UK.
- September 2025. A paper was invited for presentation at The Annals of Applied Statistics Invited Paper Session at JSM 2026.
- September 2025. Our paper You Are the Best Reviewer of Your Own Papers: The Isotonic Mechanism is accepted by Operations Research.
- September 2025. Our paper The 2020 United States Decennial Census Is More Private Than You (Might) Think is accepted by PNAS.
- September 2025. I will serve as the Scientific Integrity Chair for ICML 2026.
- August 2025. I received the Outstanding Young Talent Award in the 2025 China Annual Review of Mathematics.
- July 2025. I was selected to deliver the IMS Medallion Award and Lecture in 2027.
- May 2025. Our paper on the ICML ranking experiment will appear as a Discussion Paper in JASA.
- May 2025. I was elected as a Fellow of IMS.
- May 2025. I'll serve as the Program Chair for the ASA Text Analysis Section starting January 1, 2026.
- March 2025. We offered (with Qi Long and Inyoung Choi) a short course on large language models at ENAR 2025.
- January 2025. Our third ranking experiment (with Buxin Su) was successfully conducted at ICML 2025. See more details in OpenRank.