Papers

Journal Publications

  1. Yichun Hu, Nathan Kallus, Xiaojie Mao (2022) Fast Rates for Contextual Linear Optimization.
    • Management Science 68(6):4236-4245.
  2. Yichun Hu, Nathan Kallus, Xiaojie Mao (2022) Smooth Contextual Bandits: Bridging the Parametric and Non-differentiable Regret Regimes.
    • Operations Research (Article in Advance).
    • Preliminary version in 33rd Conference on Learning Theory (COLT), 2020.
    • Finalist, INFORMS Applied Probability Society Best Student Paper Competition, 2020.

Under Revision

  1. Yichun Hu, Nathan Kallus, Masatoshi Uehara. Fast Rates for the Regret of Offline Reinforcement Learning.
    • Minor Revision at Mathematics of Operations Research.
    • Preliminary version in 34th Conference on Learning Theory (COLT), 2021.
    • Invited for presentation at Online RL Theory Seminar.
  2. Yichun Hu, Nathan Kallus. DTR Bandit: Learning to Make Response-Adaptive Decisions with Low Regret.
    • Under 2nd-round Review after Major Revision at Journal of the American Statistical Association.

Conference Papers

  1. Mia Garrard, Hanson Wang, Benjamin Letham, Zehui Wang, Yin Huang, Yichun Hu, Chad Zhou, Norm Zhou, Eytan Bakshy (2021) Practical Policy Optimization with Personalized Experimentation.
    • NeurIPS Workshop on Causal Inference Challenges in Sequential Decision Making: Bridging Theory and Practice.



Talks

Fast Rates for Contextual Linear Optimization

  • INFORMS Annual Meeting, Indianapolis, IN 10/2022
  • Cornell ORIE Young Researchers Workshop, Ithaca, NY 10/2022
  • ORIE PhD Colloquium at Cornell Tech, New York, NY
  • INFORMS Optimization Society Conference, Greenville, SC, 03/2022

Fast Rates for the Regret of Offline Reinforcement Learning

  • RL Theory Seminar, Virtual, 11/2021 [60-min video]
  • INFORMS Annual Meeting, Anaheim, CA, 10/2021
  • 16th INFORMS Workshop on Data Mining and Decision Analytics, Anaheim, CA, 10/2021
  • 34th Annual Conference on Learning Theory (COLT 2021), Boulder, CO, 08/2021 [18-min video]

DTR Bandit: Learning to Make Response-Adaptive Decisions with Low Regret

  • INFORMS Annual Meeting, Virtual, 11/2020
  • 15th INFORMS Workshop on Data Mining and Decision Analytics, Virtual, 11/2020

Smooth Contextual Bandits: Bridging the Parametric and Non-differentiable Regret Regimes

  • 33rd Annual Conference on Learning Theory (COLT 2020), Virtual, 07/2020 [15-min video]
  • INFORMS Annual Meeting, Seattle, WA, 10/2019
  • 14th INFORMS Workshop on Data Mining and Decision Analytics, Seattle, WA, 10/2019
  • Cornell ORIE Young Researchers Workshop, Ithaca, NY, 10/2019