Sample Complexity of Asynchronous Q-Learning:
    Sharper Analysis and Variance Reduction

    Gen Li     Yuting Wei     Yuejie Chi     Yuantao Gu      Yuxin Chen
   Tsinghua      CMU         CMU        Tsinghua       Princeton

                       Abstract
     Asynchronous Q-learning aims to learn the optimal action-value function (or Q-
     function) of a Markov decision process (MDP), based on a single trajectory of
     Markovian samples induced by a behavior policy. Focusing on a γ-dis ...