Year: 2021
Type: preprint
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 that combines inverse model estimation and temporal contrastive learning to learn an abstraction that approximately satisfies these conditions. Our novel training objective is compatible with both online and offline training: it does not require a reward signal, but agents can capitalize on reward information when available. We empirically evaluate our approach on a visual gridworld domain and a set of continuous control benchmarks. Our approach learns representations that capture the underlying structure of the domain and lead to improved sample efficiency over state-of-the-art deep reinforcement learning with visual features -- often matching or exceeding the performance achieved with hand-designed compact state information.
Source: arXiv (Cornell University)
Authors: Allen, Cameron, Parikh, Neev, Gottesman, Omer, Konidaris, George
Institution: Brown University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: G2694382171, G4504108201, G4713059963, RI: Small: Collaborative Research: Hidden Parameter Markov Decision Processes: Exploiting Structure in Families of Tasks, RI: Medium: Learning Task-Specific Representations for Broadly Capable Reinforcement Learning Agents, G6671297155, CAREER: Learning Symbolic Representations for Robot Manipulation, G8105294677, G8876996369, G8929337198
OpenAlex ID: W3170931039