E-mail: zhangbh at csail dot mit dot edu
Office: 32-G638
CV, Google Scholar

I am a postdoc at MIT CSAIL, working with Profs. Constantinos Daskalakis and Gabriele Farina. I am on the academic job market!

My current research interests lie in computational game theory, especially no-regret learning in games; equilibrium computation in extensive-form games; solution concepts involving correlation, communication, and/or mediation; automated mechanism design; subgame solving; and connections to optimization. I have also done work in adversarial robustness, fairness in machine learning, and quantum computing.

I recently completed my PhD in Computer Science at Carnegie Mellon University, where I worked with Prof. Tuomas Sandholm. My thesis won an CMU SCS Dissertation Award Honorable Mention, as well as CMU nominations for the ACM and AAAI/ACM SIGAI Dissertation Awards. I was also the inaugural recipient of the CMU Hans J. Berliner Graduate Fellowship in Artificial Intelligence. Before that, I completed my undergraduate and master’s degrees in Computer Science at Stanford University, where I completed my honors thesis with Prof. Gregory Valiant.

Publications and Preprints

(αβ) denotes alphabetical ordering of authors.
* denotes equal contribution.

  2. The Complexity of Equilibrium Refinements in Potential Games
    (αβ) Ioannis Anagnostides, Maria-Florina Balcan, Kiriaki Fragkia, Tuomas Sandholm, Emanuel Tewolde, Brian Hu Zhang
    arXiv 2025

  3. Scale-Invariant Regret Matching and Online Learning with Optimal Convergence: Bridging Theory and Practice in Zero-Sum Games
    Brian Hu Zhang, Ioannis Anagnostides, Tuomas Sandholm
    arXiv 2025

  4. Decision Making under Imperfect Recall: Algorithms and Benchmarks
    Emanuel Tewolde, Brian Hu Zhang, Ioannis Anagnostides, Tuomas Sandholm, Vincent Conitzer
    UAI Workshop on Safe AI 2025

  5. General search techniques without common knowledge for imperfect-information games, and application to superhuman Fog of War chess
    Brian Hu Zhang, Tuomas Sandholm
    arXiv 2025

  6. A Polynomial-Time Algorithm for Variational Inequalities under the Minty Condition
    (αβ) Ioannis Anagnostides, Gabriele Farina, Tuomas Sandholm, Brian Hu Zhang
    arXiv 2025

  7. Learning a Game by Paying the Agents
    Brian Hu Zhang*, Tao Lin*, Yiling Chen, Tuomas Sandholm
    arXiv 2025

  8. Learning and Computation of Φ-Equilibria at the Frontier of Tractability
    Brian Hu Zhang*, Ioannis Anagnostides*, Emanuel Tewolde, Ratip Emin Berker, Gabriele Farina, Vincent Conitzer, Tuomas Sandholm
    EC 2025

  9. Expected Variational Inequalities
    Brian Hu Zhang*, Ioannis Anagnostides*, Emanuel Tewolde, Ratip Emin Berker, Gabriele Farina, Vincent Conitzer, Tuomas Sandholm
    ICML 2025 Oral Presentation

  10. Computing Game Symmetries and Equilibria That Respect Them
    Emanuel Tewolde, Brian Hu Zhang, Caspar Oesterheld, Tuomas Sandholm, Vincent Conitzer
    AAAI 2025 Oral Presentation, Best Poster Award

  11. Optimal Correlated Equilibria in General-Sum Extensive-Form Games: Fixed-Parameter Algorithms, Hardness, and Two-Sided Column-Generation
    Brian Hu Zhang, Gabriele Farina, Andrea Celli, Tuomas Sandholm
    Mathematics of Operations Research 2025
    Extended version of [11] (EC 2022)

  12. A Lower Bound on Swap Regret in Extensive-Form Games
    (αβ) Constantinos Daskalakis, Gabriele Farina, Noah Golowich, Tuomas Sandholm, Brian Hu Zhang
    arXiv 2024

  13. Efficient Φ-Regret Minimization with Low-Degree Swap Deviations in Extensive-Form Games
    Brian Hu Zhang, Ioannis Anagnostides, Gabriele Farina, Tuomas Sandholm
    NeurIPS 2024

  14. Exponential Lower Bounds on the Double Oracle Algorithm in Zero-Sum Games
    Brian Hu Zhang, Tuomas Sandholm
    IJCAI 2024

  15. Imperfect-Recall Games: Equilibrium Concepts and Their Complexity
    Emanuel Tewolde, Brian Hu Zhang, Caspar Oesterheld, Manolis Zampetakis, Tuomas Sandholm, Paul W. Goldberg, Vincent Conitzer
    IJCAI 2024

  16. Hidden-Role Games: Equilibrium Concepts and Computation
    Luca Carminati*, Brian Hu Zhang*, Gabriele Farina, Nicola Gatti, Tuomas Sandholm
    EC 2024

  17. Steering No-Regret Learners to a Desired Equilibrium
    Brian Hu Zhang*, Gabriele Farina*, Ioannis Anagnostides, Federico Cacciamani, Stephen McAleer, Andreas Haupt, Andrea Celli, Nicola Gatti, Vincent Conitzer, Tuomas Sandholm
    EC 2024

  18. Mediator Interpretation and Faster Learning Algorithms for Linear Correlated Equilibria in General Extensive-Form Games
    Brian Hu Zhang, Gabriele Farina, Tuomas Sandholm
    ICLR 2024

  19. On the Outcome Equivalence of Extensive-Form and Behavioral Correlated Equilibria
    Brian Hu Zhang, Tuomas Sandholm
    AAAI 2024

  20. Computing Optimal Equilibria and Mechanisms via Learning in Zero-Sum Extensive-Form Games
    Brian Hu Zhang*, Gabriele Farina*, Ioannis Anagnostides, Federico Cacciamani, Stephen McAleer, Andreas Haupt, Andrea Celli, Nicola Gatti, Vincent Conitzer, Tuomas Sandholm
    NeurIPS 2023

  21. Subgame Solving in Adversarial Team Games
    Brian Hu Zhang*, Luca Carminati*, Federico Cacciamani, Gabriele Farina, Pierriccardo Olivieri, Nicola Gatti, Tuomas Sandholm
    NeurIPS 2022

  22. Polynomial-Time Optimal Equilibria with a Mediator in Extensive-Form Games
    Brian Hu Zhang, Tuomas Sandholm
    NeurIPS 2022

  23. Optimal Correlated Equilibria in General-Sum Extensive-Form Games: Fixed-Parameter Algorithms, Hardness, and Two-Sided Column-Generation
    Brian Hu Zhang, Gabriele Farina, Andrea Celli, Tuomas Sandholm
    EC 2022
    Extended by [23] (Mathematics of Operation Research 2025)

  24. Team Belief DAG: Generalizing the Sequence Form to Team Games for Fast Computation of Correlated Team Max-Min Equilibria via Regret Minimization
    Brian Hu Zhang, Gabriele Farina, Tuomas Sandholm
    ICML 2023
    Earlier version in ICLR Workshop on Gamification and Multiagent Solutions (GMAS) 2022

  25. Polynomial-Time Sum-of-Squares Can Robustly Estimate Mean and Covariance of Gaussians Optimally
    (αβ) Pravesh K. Kothari, Peter Manohar, Brian Hu Zhang
    ALT 2022

  26. Team Correlated Equilibria in Zero-Sum Extensive-Form Games via Tree Decompositions
    Brian Hu Zhang, Tuomas Sandholm
    AAAI 2022

  27. Subgame solving without common knowledge
    Brian Hu Zhang, Tuomas Sandholm
    NeurIPS 2021 Spotlight
    AAAI Workshop on Reinforcement Learning in Games 2022 Oral Presentation

  28. Finding and Certifying (Near-)Optimal Strategies in Black-Box Extensive-Form Games
    Brian Hu Zhang, Tuomas Sandholm
    AAAI 2021
    AAAI Workshop on Reinforcement Learning in Games 2021 Oral Presentation

  29. Small Nash Equilibrium Certificates in Very Large Games
    Brian Hu Zhang, Tuomas Sandholm
    NeurIPS 2020

  30. Sparsified Linear Programming for Zero-Sum Equilibrium Finding
    Brian Hu Zhang, Tuomas Sandholm
    ICML 2020

  31. A Spectral View of Adversarially Robust Features
    Shivam Garg, Vatsal Sharan*, Brian Hu Zhang*, Gregory Valiant
    NeurIPS 2018 Spotlight

  32. Mitigating Unwanted Biases with Adversarial Learning
    Brian Hu Zhang, Blake Lemoine, Margaret Mitchell
    AIES 2018

  33. Advantages of Unfair Quantum Ground-State Sampling
    Brian Hu Zhang, Gene Wagenbreth, Victor Martin-Mayor, Itay Hen
    Nature Scientific Reports 2016

Teaching

Carnegie Mellon University

Stanford University