Representation of Reinforcement Learning Policies in Reproducing Kernel Hilbert Spaces

TL;DR

This paper introduces a framework for representing reinforcement learning policies within reproducing kernel Hilbert spaces, enabling low-dimensional embeddings with strong theoretical guarantees and robust performance.

Contribution

It presents a novel RKHS-based policy representation method that provides theoretical guarantees and maintains high return performance in RL tasks.

Findings

Policies can be embedded in low-dimensional RKHS space

Embedded policies incur minimal loss in expected return

Method demonstrates robustness across RL domains

Abstract

We propose a general framework for policy representation for reinforcement learning tasks. This framework involves finding a low-dimensional embedding of the policy on a reproducing kernel Hilbert space (RKHS). The usage of RKHS based methods allows us to derive strong theoretical guarantees on the expected return of the reconstructed policy. Such guarantees are typically lacking in black-box models, but are very desirable in tasks requiring stability. We conduct several experiments on classic RL domains. The results confirm that the policies can be robustly embedded in a low-dimensional space while the embedded policy incurs almost no decrease in return.