Abstract:In this paper, we present a stochastic convergence proof, under suitable conditions, of a certain class of actor-critic algorithms for finding approximate solutions to entropy-regularized MDPs using the machinery of stochastic approximation. To obtain this overall result, we provide three fundamental results that are all of both practical and theoretical interest: we prove the convergence of policy evaluation with general regularizers when using linear approximation architectures, we derive an entropy-regularized policy gradient theorem, and we show convergence of entropy-regularized policy improvement. We also provide a simple, illustrative empirical study corroborating our theoretical results. To the best of our knowledge, this is the first time such results have been provided for approximate solution methods for regularized MDPs.
Abstract:This paper extends off-policy reinforcement learning to the multi-agent case in which a set of networked agents communicating with their neighbors according to a time-varying graph collaboratively evaluates and improves a target policy while following a distinct behavior policy. To this end, the paper develops a multi-agent version of emphatic temporal difference learning for off-policy policy evaluation, and proves convergence under linear function approximation. The paper then leverages this result, in conjunction with a novel multi-agent off-policy policy gradient theorem and recent work in both multi-agent on-policy and single-agent off-policy actor-critic methods, to develop and give convergence guarantees for a new multi-agent off-policy actor-critic algorithm.