Vincent Y. F. Tan 陈延福 (he/him • 他)

Professor Department of Mathematics & Department of Electrical and Computer Engineering (ECE) National University of Singapore (NUS) NRF Investigator Deputy Head (Graduate Programmes) Affiliations IORA, IDS, ISEP Office: S17-05-20 (Mathematics) and E4-06-06 (ECE) Phone: (65) 6516-2133 Email: vtan @ nus. edu. sg My Curriculum Vitae. My Google Scholar Profile. My Erdős number is 3 (Tan -> Willsky -> Bollobás -> Erdős).

Featured

• Recruiting Postdoctoral Researchers. Applications are invited to join my NRF Investigatorship on information-theoretic foundations of large language models. Please see the advertisement for more details.


• Teaching This Semester. Information-Theoretic Methods in Statistical Learning (MA6225). Thursdays, 10:00–13:00, S17-05-11. Open to all NUS students.

News (Archive)

• Jul 2026: Heartiest congratulations to research fellow Yihang Gao for being offered a tenure track Assistant Professor position at Hong Kong Baptist University (HKBU) Math department.


• Jul 2026: Posted a preprint on the arxiv "Statistical Inference for Rank Allocation in Low-Rank Adaptation". Joint work with Yihang Gao.


• Jul 2026: Ping Li and Canzhe Zhao join the group as a Research Fellows. Welcome!


• Jul 2026: Paper "Finite-Time Minimax Bounds and an Optimal Lyapunov Policy in Queueing Control" accepted to Operations Research. Joint work with Yujie Liu and Yunbei Xu.


• Jul 2026: Three papers to be presented at the 19th European Workshop on Reinforcement Learning (EWRL 2026). (i) "Best Arm Identification with Minimal Regret" (with Junwen Yang and Tianyuan Jin); (ii) "Reinforcement Learning with Abstention: Interaction-Aware Regret Bounds" (with Yuan Cheng); (iii) "On the Benefits of Free Exploration for Regret Minimization in Multi-Armed Bandits" (with Yunlong Hou and Zixin Zhong).


• Jul 2026: Posted a preprint on the arxiv "Balancing Expressivity and Learnability in Quantum Kernel Bandit Optimization". Joint work with Ph.D. student Yuqi Huang and collaborator Sharu Theresa Jose. To be presented at UAI 2026.


• Jun 2026: Posted a preprint on the arxiv "Bayesian Best-Arm Identification with Abstention: A Polynomial-to-Exponential Phase Transition". Joint work with Ph.D. student Yuqi Huang and research fellow Yunlong Hou.


• Jun 2026: Reappointed to the University Promotion and Tenure Committee (UPTC) and the University Teaching Excellence Committee (UTEC). Taking up the role of Deputy Head (Graduate Programmes) in the Department of Mathematics in the new academic year.


• Jun 2026: Yulian Wu joins the group as a Research Fellow. Welcome!


• Jun 2026: Hosted a visit by Dr. Yuko Kuroki.


• Jun 2026: Rated 4.8/5.0 and 4.9/5.0 respectively for my teaching of EE5138 Optimization for ECE and EE6138 Optimization for ECE (Advanced) last semester. See some review quotes here and here. I really enjoyed teaching this course!


• Jun 2026: Two papers accepted to UAI 2026: (i) Balancing Expressivity and Learnability in Quantum Kernel Bandit Optimization (joint work with Ph.D. student Yuqi Huang and collaborator Sharu Theresa Jose) and (ii) Dynamic Regret in Outlier-Oblivious Online Optimization using Nonconvex Robust Losses (joint work with former research fellows Adarsh Barik and Anand Krishna).


• May 2026: Posted a paper on the arxiv: "On the Benefits of Free Exploration for Regret Minimization in Multi-Armed Bandits". Joint work with Yunlong Hou and Zixin Zhong.


• May 2026: Posted two papers on the arxiv: (i) "Bandit Convex Optimization with Gradient Prediction Adaptivity" (joint work with Shuche Wang and Adarsh Barik) and (ii) "How Many Different Outputs Can a Transformer Generate?" (joint work with Maxime Meyer, Mario Michelessa and Caroline Chaux) which is to be presented at ICML as an oral presentation (and spotlight poster).


• May 2026: Paper "Avoiding exp(k^*) Scaling for Thompson Sampling in Combinatorial Semi-Bandits: From Multiple Seeds to a Single Seed" accepted to the Conference on Learning Theory (COLT). Joint work with Research Fellow Tianyuan Jin and others.


• May 2026: Paper "How Many Different Outputs Can a Transformer Generate?" accepted to the International Conference on Machine Learning (ICML) as an Oral Presentation and Spotlight Poster (top 2.2% of submissions). Joint work with Ph.D. student Maxime Meyer, collaborator Mario Michelessa and colleague Caroline Chaux.


• Apr 2026: Awarded an NRF Investigatorship (Class of 2026) for the proposal titled "Fundamental Limits of Language Models via Information Theory" (Total Project Cost ≈ S$2.85 million). I am looking to hire multiple postdoctoral scholars and PhD students. Please see this advertisement.


• Apr 2026: Paper "Error Analyses of Auto-Regressive Video Diffusion Models: A Unified Framework" accepted to the Journal of Machine Learning Research (JMLR). Joint work with former Ph.D. student Fengzhuo Zhang and others.


• Apr 2026: Paper "Best Arm Identification with Minimal Regret" accepted to the Journal of Machine Learning Research (JMLR). Joint work with Junwen Yang and Tianyuan Jin.


• Mar 2026: Paper "Almost Asymptotically Optimal Active Clustering Through Pairwise Observations" accepted to ISIT 2026! Joint work with Rachel Teo, P. N. Karthik and Ramya Vinayak.


• Mar 2026: Posted a paper on the arxiv "Timely Best Arm Identification in Restless Shared Networks". Joint work with Mengqiu Zhou and Meng Zhang.


• Mar 2026: Paper "Age-Optimal Best Arm Identification" accepted to WiOpt 2026, the 24th International Symposium on Modeling and Optimization in Mobile, Ad hoc, and Wireless Networks. Joint work with Mengqiu Zhou, Le Yang and Meng Zhang.


• Mar 2026: Paper "Multi-Armed Bandits with Abstention" accepted to the Transactions on Machine Learning Research (TMLR). Joint work with Junwen Yang and Tianyuan Jin.


• Mar 2026: Congrats to PhD student Shuche Wang for successfully defending his excellent thesis.


• Mar 2026: Congratulations to former Ph.D. student Jingyi Zhao for being offered a tenure track Assistant Professor position in Business Analytics and Decision Science at Vienna University of Economics and Business.

Research Interests

Online Decision Making, Multi-Armed Bandits, Reinforcement Learning
Information Theory with Applications to Machine Learning
Statistical Signal Processing

Selected Recent Publications

1. Best Arm Identification with Minimal Regret
   Junwen Yang, Vincent Y. F. Tan, and Tianyuan Jin
   Journal of Machine Learning Research, Vol. 27, No. 86, Pages 1 - 41, 2026

2. Asymptotically and Minimax Optimal Regret Bounds for Multi-Armed Bandits with Abstention
   Junwen Yang, Tianyuan Jin, and Vincent Y. F. Tan
   Transactions on Machine Learning Research, Apr 2026

3. On the Exponential Convergence for Offline RLHF with Pairwise Comparisons
   Zhirui Chen and Vincent Y. F. Tan
   Proc. of the 40th AAAI Conference on Artificial Intelligence (AAAI) (Alignment Track), Singapore, Feb 2026 (Oral Presentation, AR: ≈ 22.7%)

4. Parameter-free Algorithms for the Stochastically Extended Adversarial Model
   Shuche Wang, Adarsh Barik, Peng Zhao, and Vincent Y. F. Tan
   Proc. of the 39th Annual Conference on Neural Information Processing Systems (NeurIPS), San Diego, CA, Dec 2025 (AR ≈ 24.5%)

5. Ensemble-Tight Second-Order Asymptotics and Exponents for Guessing-Based Decoding with Abandonment
   Vincent Y. F. Tan and Hamdi Joudeh
   IEEE Transactions on Information Theory, Vol. 71, No. 10, Pages 7555 - 7567, Oct 2025

Selected Older Publications

1. Common Information, Noise Stability, and Their Extensions
   Lei Yu and Vincent Y. F. Tan
   Foundations and Trends® in Communications and Information Theory, Vol. 19, No. 2, Pages 107 - 389, 2022

2. Asymptotic Estimates in Information Theory with Non-Vanishing Error Probabilities
   Vincent Y. F. Tan
   Foundations and Trends® in Communications and Information Theory, Vol. 11, Nos. 1-2, Pages 1 – 184, 2014

3. Time-Division is Optimal for Covert Communication over Some Broadcast Channels [Slides]
   Vincent Y. F. Tan and Si-Hyeon Lee
   IEEE Transactions on Information Forensics and Security, Vol. 14, No. 5, Pages 1377 - 1389, May 2019

4. Automatic Relevance Determination in Nonnegative Matrix Factorization with the β-Divergence [Slides] [Code]
   Vincent Y. F. Tan and Cédric Févotte
   IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 35, No. 7, Pages 1592 - 1605, Jul 2013

5. A Large-Deviation Analysis of the Maximum-Likelihood Learning of Markov Tree Structures [Slides]
   Vincent Y. F. Tan, Animashree Anandkumar, Lang Tong and Alan S. Willsky
   IEEE Transactions on Information Theory, Vol. 57, No. 3, Pages 1714 - 1735, Mar 2011