Towards General Function Approximation in Zero-Sum Markov Games

Authors: Baihe Huang, Jason D. Lee, Zhaoran Wang, Zhuoran Yang

ICLR 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical This paper considers two-player zero-sum finite-horizon Markov games with simultaneous moves. The study focuses on the challenging settings where the value function or the model is parameterized by general function classes. Provably efficient algorithms for both decoupled and coordinated settings are developed.
Researcher Affiliation Academia Baihe Huang School of Mathematical Sciences Peking University baihehuang@pku.edu.cn Jason D. Lee Department of Electrical and Computer Engineering Princeton University Jasondl@princeton.edu Zhaoran Wang Departments of Industrial Engineering & Management Sciences Northwestern University zhaoranwang@gmail.com Zhuoran Yang Department of Statistics and Data Science Yale University zhuoran.yang@yale.edu
Pseudocode Yes Algorithm 1 Optimistic Nash Elimination for Markov Games (ONEMG)
Open Source Code No The paper does not contain any statement about releasing source code or a link to a code repository.
Open Datasets No The paper is theoretical and focuses on algorithms and sample complexity bounds, rather than reporting experiments on specific publicly available datasets.
Dataset Splits No The paper is theoretical and does not discuss dataset splits for experimental reproduction.
Hardware Specification No The paper is theoretical and does not describe any specific hardware used for running experiments.
Software Dependencies No The paper is theoretical and does not mention specific software dependencies with version numbers for experimental reproducibility.
Experiment Setup No The paper is theoretical and does not describe a concrete experimental setup with hyperparameters or system-level training settings.