Qinghua Liu

I am a Ph.D. candidate in the Department of Electrical and Computer Engineering at Princeton University. I am fortunate to be advised by Chi Jin. My current research interests focus on reinforcement learning theory. I am particularly interested in RL problems related to function approximation, partial observability and game theory. Previously, I received a B.E. degree in Electrical Engineering and a B.S. degree in Mathematics from Tsinghua University in 2018.

Recent Papers

(α-β order) denotes alphabetical authorship ordering

• Sample-Efficient Reinforcement Learning of Partially Observable Markov Games
  Qinghua Liu, Csaba Szepesvári, Chi Jin
  arXiv preprint; preliminary version appeared in 2022 European Workshop on Reinforcement Learning (Oral presentation)
  

• Policy Optimization for Markov Games: Unified Framework and Faster Convergence
  Runyu Zhang, Qinghua Liu, Huan Wang, Caiming Xiong, Na Li, Yu Bai
  arXiv preprint

• When Is Partially Observable Reinforcement Learning Not Scary?
  Qinghua Liu, Alan Chung, Csaba Szepesvári, Chi Jin
  COLT, 2022

• Learning Markov Games with Adversarial Opponents: Efficient Algorithms and Fundamental Limits
  Qinghua Liu, Yuanhao Wang, Chi Jin
  ICML, 2022 (Long presentation)

• A Deep Reinforcement Learning Approach for Finding Non-Exploitable Strategies in Two-Player Atari Games
  Zihan Ding, Dijia Su, Qinghua Liu, Chi Jin
  arXiv preprint

• V-Learning – A Simple, Efficient, Decentralized Algorithm for Multiagent RL
  (α-β order) Chi Jin, Qinghua Liu, Yuanhao Wang, Tiancheng Yu
  arXiv preprint; Best Paper in ICLR 2022 Workshop on Gamification and Multiagent Solutions
  

• The Power of Exploiter: Provable Multi-Agent RL in Large State Spaces
  (α-β order) Chi Jin, Qinghua Liu, Tiancheng Yu
  ICML, 2022; preliminary version appeared in ICML 2021 Workshop on Reinforcement Learning Theory
  

• Bellman Eluder Dimension: New Rich Classes of RL Problems, and Sample-Efficient Algorithms [Slides] [RL Theory Seminar]
  (α-β order) Chi Jin, Qinghua Liu, Sobhan Miryoosefi
  NeurIPS, 2021 (Spotlight)

• A Sharp Analysis of Model-based Reinforcement Learning with Self-Play
  Qinghua Liu, Tiancheng Yu, Yu Bai, Chi Jin
  ICML, 2021

• Provable Rich Observation Reinforcement Learning with Combinatorial Latent States
  Dipendra Misra, Qinghua Liu, Chi Jin, John Langford
  ICLR, 2021

• Sample-Efficient Reinforcement Learning of Undercomplete POMDPs [Slides] [RL Theory Seminar]
  (α-β order) Chi Jin, Sham M. Kakade, Akshay Krishnamurthy, Qinghua Liu
  NeurIPS, 2020 (Spotlight)

• Tackling the Objective Inconsistency Problem in Heterogeneous Federated Optimization
  Jianyu Wang, Qinghua Liu, Hao Liang, Gauri Joshi, H. Vincent Poor
  NeurIPS, 2020; full version in IEEE Transactions on Signal Processing

Technical Notes

• A Tight Lower Bound for Uniformly Stable Algorithms
  (α-β order) Qinghua Liu, Zhou Lu
  arXiv preprint

Services

Reviewers for COLT, NeurIPS, ICML, JMLR, ICLR, ALT, IEEE-TSP, IEEE-TAC, etc.