Reinforcement Learning with Function Approximation for Non-Markov Processes

by @math-papers

Introduction Model-free reinforcement learning methods aim to compute approximately optimal control policies, or the value function of a stochastic control problem, directly from interaction data without constructing a m…