Learning Markov State Abstractions for Deep Reinforcement Learning

arXiv:2106.04379v4 Announce Type: replace
Abstract: A fundamental assumption of reinforcement learning in Markov decision processes (MDPs) is that the relevant decision process is, in fact, Markov. However, when MDPs have rich observations, agents typically learn by way of an abstract state representation, and such representations are not guaranteed to preserve the Markov property. We introduce a novel set of conditions and prove that they are sufficient for learning a Markov abstract state representation. We then describe a practical training procedure …

abstract abstractions agents arxiv cs.ai cs.lg decision however learn markov process processes reinforcement reinforcement learning representation state stat.ml type