Buxin Su
I am a Ph.D. student in Applied Mathematics and Computational Science at the University of Pennsylvania, where I have worked with Weijie Su and Bingxin Zhao. Before that, I received an M.S. in Pure Mathematics from Penn and a B.S. in both Pure Mathematics and Applied Mathematics from the University of Toronto, graduating with High Distinction.
My research interests primarily lie in privacy-preserving and trustworthy data science, with applications to machine learning and the social sciences. I am particularly interested in improving the quality and reliability of peer review at AI/ML conferences. Our recent work on using author–reviewer misalignment to flag unreliable or miscalibrated reviews has been incorporated into the ICML 2026 review process.
Recent & Featured Papers
A selection of recent work — see Publications for the full list.
-
The ICML 2023 Ranking Experiment: Examining Author Self-Assessment in ML/AI Peer ReviewJournal of the American Statistical Association, 2025
Introducing self-ranking to ICML 2026 -
The 2020 United States Decennial Census Is More Private Than You (Might) ThinkProceedings of the National Academy of Sciences, 2025
Featured by New Scientist -
How to Find Fantastic AI Papers: Self-Rankings as a Powerful Predictor of Scientific Impact Beyond Peer ReviewNature Computational Science, 2026
Featured by Nature Index
News & Awards
- Jul 2026 “How to Find Fantastic AI Papers” is published in Nature Computational Science and featured by Nature Index and the Wharton AI & Analytics Initiative.
- May 2026 Received the NESS Student Research Awards at the New England Statistics Symposium.
- Jan 2026 Named a finalist for Two Sigma’s PhD Fellowship and gave an invited talk at the finalist reception.
- Jan 2026 ICML 2026 introduced a self-ranking policy for reviews, building on our research. I am also serving as an Associate Chair for ICML 2026.
- Oct 2025 Received the Lawrence D. Brown Best Paper Award from the Wharton School, University of Pennsylvania.
- Sep 2025 “The 2020 United States Decennial Census Is More Private Than You (Might) Think” is published in PNAS and featured by New Scientist.
- Aug 2025 “The ICML 2023 Ranking Experiment” is published in JASA and selected as a Discussion Paper (equivalent to best paper award).
- Jul 2025 Received the ICSA NSF Travel Award at the ICSA 2025 Applied Statistics Symposium.
- Jan 2024 Won the Student Best Paper Award, Statistical Learning and Data Science Section, American Statistical Association.