Yunbei Xu (许云贝)

Email: yunbei [at] nus.edu.sg

I am a Presidential Young Professor in the Department of Industrial Systems Engineering & Management at the National University of Singapore. 

My research interests include machine learning, operations research, and statistical physics. 

I am looking for motivated PhD students to join my research group.


Background

Postdoc, Massachusetts Institute of Technology

   Laboratory for Information and Decision Systems

   Advisor: Sasha Rakhlin

Doctor of Philosophy, Columbia University

   Graduate School of Business

   Advisor: Assaf Zeevi

Bachelor of Science, Peking University

   Department of Pure Mathematics

Research

•  Assouad, Fano, and Le Cam with Interaction: A Unifying Lower Bound Framework and Characterization for Bandit Learnability

   with Fan Chen, Dylan J Foster, Yanjun Han, Jian Qian, and Alexander Rakhlin. 

   - Conference on Neural Information Processing Systems (NeurIPS) 2024.

   - Spotlight

•  Statistical Properties of Robust Satisficing

   with Zhiyi Li and Ruohan Zhan. 

   - International Conference on Machine Learning (ICML) 2024.

•  Bayesian Design Principles for Frequentist Sequential Learning (this is the arXiv link to the full version, which is the authors’ recommended version)

   with Assaf Zeevi. 

   - short version in International Conference on Machine Learning (ICML) 2023. Code

   - ICML Outstanding Paper Award

   - INFORMS George Nicholson Student Paper Competition, First Place

   - Applied Probability Society Best Student Paper Award, Finalist

•  Towards Optimal Problem Dependent Generalization Error Bounds in Statistical Learning Theory

   with Assaf Zeevi. 

   - Mathematics of Operations Research, 2024.

   - Applied Probability Society Best Student Paper Award, Finalist

   Towards Problem-dependent Optimal Learning Rates

   with Assaf Zeevi.

   - Conference on Neural Information Processing Systems (NeurIPS) 2020.

   - Spotlight (top 2.9%)

•  Upper Counterfactual Confidence Bounds: a New Optimism Principle for Contextual Bandits

   with Assaf Zeevi.

•  Acceleration of Primal-Dual Methods by Preconditioning and Simple Subproblem Procedures 

   with Yanli Liu and Wotao Yin.

   - Journal of Scientific Computing, 2021. Code

Teaching

Teaching Assistant: Statistical Physics, Markets and Algorithms (Fall 2019, instructed by Yash Kanoria)