About Me


I am an assistant professor in Computer Science and Data Science Institute at UChicago, and direct the Sigma Lab (Strategic IntelliGence for Machine Agents). I also spend time at Google Research as a part time research scientist.


Prior to UChicago, I was an assistant professor at UVA and (even before) a postdoc at Harvard hosted by Yiling Chen and David Parkes. I received a PhD in Computer Science from USC advised by Shaddin Dughmi and Milind Tambe (now at Harvard), after writing this dissertation which was recognized by the ACM SIGecom Dissertation Award and IFAAMAS Victor Lesser Distinguished Dissertation Award.


Interested in joining our group? Please read this.


Contact


Office: Crerar 260

E-mail: haifengxu AT uchicago DOT edu

We are looking for PhD students for 2024-2025 academic year to work on agentic AI (particularly on incentives and persuasion for LLMs), from both CS PhD and DSI PhD programs. It suffices to mention my name in your application -- there is no need to additionally email me as it won't affect the admission chance.


About My Research


I work on the economics of machine learning, such as understanding the economic aspects of machine learning itself and designing ML algorithms for economic problems. The following are a few research themes of my recent focus (also see my survey paper). More broadly, I am interested in decision making and machine learning in multi-agent setups, particularly in informationally complex environments with limited or asymmetric access to information. Please see our Sigma Research Lab for more details.

  1. Economics for data and machine learning (e.g., acquiring/eliciting/selling data and ML)

    2. Representative Recent Papers (see also my tutorial at AAAI23):

    1. Mechanism Design for Large Language Models

      (α-β) Paul Duetting, Vahab Mirrokni, Renato Paes Leme, Haifeng Xu, and Song Zuo

      WWW 2024: Proc. ACM Web Conference, 2024 (Best Paper Award)

      Selected as Highlights Beyond EC at EC 2024


    2. A Truth Serum for Eliciting Self-Evaluations in Scientific Reviews

      (Randomized order) Jibang Wu, Haifeng Xu, Yifan Guo, Weijie Su

      working paper


    3. Optimal Pricing of Information

      (α-β) Shuze Liu, Weiran Shen and Haifeng Xu

      EC 2021: Proc. 22th ACM Conference on Economics and Computation, 2021.



  2. The dynamics and economy of content creation and recommendation

    3. Representative Recent Papers (see also my talk at Google ML seminar):

    1. Rethinking Incentives in Recommender Systems: Are Monotone Rewards Always Beneficial?

      Fan Yao, Chuanhao Li, Karthik Abinav Sankararaman, Yiming Liao, Yan Zhu, Qifan Wang, Hongning Wang and Haifeng Xu

      NeurIPS 2023: Proc. 37th Conference on Neural Information Processing Systems, 2023.


    2. Follow-ups Also Matter: Improving Contextual Bandits via Post-serving Contexts

      Chaoqi Wang, Ziyu Ye, Zhe Feng, Ashwinkumar Badanidiyuru and Haifeng Xu

      NeurIPS 2023: Proc. 37th Conference on Neural Information Processing Systems, 2023 (spotlight)


    3. How Bad is Top-K Recommendation under Competing Content Creators?

      Fan Yao, Chuanhao Li, Denis Nekipelov, Hongning Wang and Haifeng Xu

      ICML 2023: Proc. 40th International Conference on Machine Learning, 2023 (live presentation)


  3. Learning in strategic/economic setups (e.g., learning game-theoretic decisions)

    4. Representative Recent Papers (see also a recent talk I gave):

    1. Multi-Agent Learning for Iterative Dominance Elimination: Formal Barriers and New Algorithms

      (α-β) Jibang Wu, Haifeng Xu and Fan Yao

      COLT 2022: Proc. 35th Annual Conference on Learning Theory, 2022.


    2. PAC-Learning for Strategic Classification

      (α-β) Ravi Sundaram, Anil Vullikanti, Haifeng Xu, Fan Yao

      The Journal of Machine Learning Research (JMLR) (also appeared at ICML 2021 as long oral, 3%)


    3. Learning to Persuade on the Fly: Robustness Against Ignorance

      You Zu, Krishnamurthy Iyer, Haifeng Xu

      Operations Research, to appear (also appears at EC 2021).


  4. Algorithmic foundation and new frontiers of optimal information design

    5. Representative Recent Papers (see also my tutorial at EC23):

    1. Algorithmic Cheap Talk

      (α-β) Yakov Babichenko, Inbal Talgam-Cohen, Haifeng Xu and Konstantin Zabarnyi

      EC 2024: Proc. 25th ACM Conference on Economics and Computation, 2024.


    2. Sequential Information Design: Markov Persuasion Process and Its Efficient Reinforcement Learning

      Jibang Wu, Zixuan Zhang, Zhe Feng, Zhaoran Wang, Zhuoran Yang, Michael I. Jordan and Haifeng Xu

      EC 2022: Proc. 23th ACM Conference on Economics and Computation, 2022.


    3. Algorithmic Bayesian Persuasion

      (α-β) Shaddin Dughmi, Haifeng Xu

      SICOMP: SIAM Journal on Computing (Invited to special issue for STOC 2016)



Recent News

PHP Hits Count