Provably Correct Optimization and Exploration with Non-linear Policies

                  Fei Feng 1 Wotao Yin 1 Alekh Agarwal 2 Lin Yang 3

             Abstract                rer & Geist, 2014; Geist et al., 2019; Abbasi-Yadkori et al.,
    Policy optimization methods remain a powerful       2019; Agarwal et al., 2020c; Bhandari & Russo, 2019) when
    workhorse in empirical Reinforcement Learning       the agent has access to a distribution over states which is
    (RL), with a focus on neur ...