Gen Li

I am a postdoctoral researcher in the Department of Statistics and Data Science at the Wharton School at University of Pennsylvania, co-advised by Professor Yuxin Chen and Professor Yuting Wei. Previously, I received my Ph.D. in the Department of Electronic Engineering at Tsinghua University, advised by Professor Yuantao Gu. Before my Ph.D., I received my bachelor's degree from the Department of Electronic Engineering and Department of Mathematics at Tsinghua University in 2016. I am interested in reinforcement learning, high-dimensional statistics, machine learning, signal processing, and mathematical optimization. E-mail: ligen [AT] wharton [DOT] upenn [DOT] edu Google Scholar CV

Recent Papers

• Is Q-Learning Minimax Optimal? A Tight Sample Complexity Analysis
  G. Li, C. Cai,Y. Chen, Y. Wei, Y. Chi
  Operations Research 2023
  

• Breaking the Sample Size Barrier in Model-Based Reinforcement Learning with a Generative Model
  G. Li, Y. Wei, Y. Chi, Y. Gu, Y. Chen
  Operations Research 2023, accepted in part to NeurIPS 2020
  

• Towards Understanding Variation-Constrained Deep Neural Networks
  G. Li, J. Ding
  IEEE Transactions on Signal Processing 2023
  

• Approximate message passing from random initialization with applications to $Z_2$ synchronization
  G. Li, W. Fan, Y. Wei
  

• Fast Computation of Optimal Transport via Entropy-Regularized Extragradient Methods
  G. Li, Y. Chen, Y. Chi, H. V. Poor, Y. Chen
  

• Softmax Policy Gradient Methods Can Take Exponential Time to Converge
  G. Li, Y. Wei, Y. Chi, Y. Chen
  Mathematical Programming 2022, accepted in part to COLT 2021
  

• Minimax-Optimal Multi-Agent RL in Markov Games With a Generative Model
  G. Li, Y. Chi, Y. Wei, Y. Chen
  Neural Information Processing Systems (NeurIPS) 2022, Oral presentation
  

• A Non-Asymptotic Framework for Approximate Message Passing in Spiked Models
  G. Li, Y. Wei
  

• Settling the sample complexity of model-based offline reinforcement learning
  G. Li, L. Shi, Y. Chen, Y. Chi, Y. Wei
  

• Breaking the Sample Complexity Barrier to Regret-Optimal Model-free Reinforcement Learning
  G. Li, L. Shi, Y. Chen, Y. Gu, Y. Chi
  Information and Inference: A Journal of the IMA 2022, accepted in part to NeurIPS 2021, Spotlight presentation
  

• Sample Complexity of Asynchronous Q-Learning: Sharper Analysis and Variance Reduction
  G. Li, Y. Wei, Y. Chi, Y. Gu, Y. Chen
  IEEE Transactions on Information Theory 2022, accepted in part to NeurIPS 2020
  

• Model-Based Reinforcement Learning Is Minimax-Optimal for Offline Zero-Sum Markov Games
  Y. Yan, G. Li, Y. Chen, J. Fan
  

• The Efficacy of Pessimism in Asynchronous Q-Learning
  Y. Yan, G. Li, Y. Chen, J. Fan
  

• Pessimistic Q-Learning for Offline Reinforcement Learning: Towards Optimal Sample Complexity
  L. Shi, G. Li, Y. Wei, Y. Chen, Y. Chi
  International Conference on Machine Learning (ICML) 2022
  

• Nonconvex Low-Rank Tensor Completion from Noisy Data
  C. Cai, G. Li, H. V. Poor, and Y. Chen
  Operations Research 2022
  

• Sample-Efficient Reinforcement Learning Is Feasible for Linearly Realizable MDPs with Limited Revisiting
  G. Li, Y. Chen, Y. Chi, Y. Gu, Y. Wei
  Neural Information Processing Systems (NeurIPS) 2021
  

• Minimax Estimation of Linear Functions of Eigenvectors in the Face of Small Eigen-Gaps
  G. Li, C. Cai, Y. Gu, H. V. Poor, Y. Chen
  

• Softmax Policy Gradient Methods Can Take Exponential Time to Converge
  G. Li, Y. Wei, Y. Chi, Y. Gu, Y. Chen
  Conference on Learning Theory (COLT) 2021
  

• Tightening the Dependence on Horizon in the Sample Complexity of Q-Learning
  G. Li, C. Cai,Y. Chen, Y. Gu, Y. Wei, Y. Chi
  International Conference on Machine Learning (ICML) 2021
  

• Theory of spectral method for union of subspaces-based random geometry graph
  G. Li, Y. Gu
  International Conference on Machine Learning (ICML) 2021
  

• Subspace Estimation from Unbalanced and Incomplete Data Matrices: $\ell_{2, infty}$ Statistical Guarantees
  C. Cai, G. Li, Y. Chi, H. V. Poor, and Y. Chen
  Annals of Statistics 2021
  

• Rigorous Restricted Isometry Property for Low-Dimensional Subspaces
  G. Li, Q. Liu, Y. Gu
  Applied and Computational Harmonic Analysis 2020
  

• Lower Bound for RIP Constants and Concentration of Sum of Top Order Statistics
  G. Li, X. Xu, Y. Gu
  IEEE Transactions on Signal Processing 2020
  

• Phase transitions of spectral initialization for high-dimensional non-convex estimation
  Y. M. Lu, G. Li
  Information and Inference: A Journal of the IMA 2020
  

• Unraveling the Veil of Subspace RIP Through Near-Isometry on Subspaces
  X. Xu, G. Li, Y. Gu
  IEEE Transactions on Signal Processing 2020