Combining Reinforcement Learning and Tensor Networks, with an Application to Dynamical Large Deviations

arXiv:2209.14089v2 Announce Type: replace-cross
Abstract: We present a framework to integrate tensor network (TN) methods with reinforcement learning (RL) for solving dynamical optimisation tasks. We consider the RL actor-critic method, a model-free approach for solving RL problems, and introduce TNs as the approximators for its policy and value functions. Our "actor-critic with tensor networks" (ACTeN) method is especially well suited to problems with large and factorisable state and action spaces. As an illustration of the applicability of ACTeN we solve …

abstract actor actor-critic application arxiv cond-mat.stat-mech cs.lg framework free network networks optimisation policy reinforcement reinforcement learning tasks tensor type