Bootstrapped Representations in Reinforcement Learning

TL;DR

This paper provides a theoretical analysis of state representations learned by bootstrapping methods in reinforcement learning, compares them with other approaches, and introduces new auxiliary learning rules validated on classic domains.

Contribution

It offers the first theoretical characterization of representations learned by temporal difference learning and designs new auxiliary rules based on this analysis.

Findings

TD learning captures different features than Monte Carlo methods.

Auxiliary rules improve policy evaluation performance.

Empirical results on classic domains validate the proposed methods.

Abstract

In reinforcement learning (RL), state representations are key to dealing with large or continuous state spaces. While one of the promises of deep learning algorithms is to automatically construct features well-tuned for the task they try to solve, such a representation might not emerge from end-to-end training of deep RL agents. To mitigate this issue, auxiliary objectives are often incorporated into the learning process and help shape the learnt state representation. Bootstrapping methods are today's method of choice to make these additional predictions. Yet, it is unclear which features these algorithms capture and how they relate to those from other auxiliary-task-based approaches. In this paper, we address this gap and provide a theoretical characterization of the state representation learnt by temporal difference learning (Sutton, 1988). Surprisingly, we find that this representation differs from the features learned by Monte Carlo and residual gradient algorithms for most transition structures of the environment in the policy evaluation setting. We describe the efficacy of these representations for policy evaluation, and use our theoretical analysis to design new auxiliary learning rules. We complement our theoretical results with an empirical comparison of these learning rules for different cumulant functions on classic domains such as the four-room domain (Sutton et al, 1999) and Mountain Car (Moore, 1990).