On Reward-Free RL with Kernel and Neural Function Approximations:
           Single-Agent MDP and Markov Game

                Shuang Qiu 1 Jieping Ye 1 Zhaoran Wang 2 Zhuoran Yang 3

              Abstract                is large and function approximators such as neural networks
    To achieve sample efficiency in reinforcement        are employed. To achieve sample efficiency, any RL algo-
    learning (RL), it necessitates to efficiently explore    rithm needs to accurately learn the transit ...