A Deep Reinforcement Learning Approach to Marginalized Importance Sampling with the Successor Representation

TL;DR

This paper introduces a deep reinforcement learning method that computes density ratios for off-policy evaluation using successor representations, enabling stable learning in complex environments.

Contribution

It bridges marginalized importance sampling with deep RL by leveraging successor representations, simplifying optimization, and improving applicability to high-dimensional problems.

Findings

Effective in Atari and MuJoCo environments

Stable and scalable to high-dimensional domains

Outperforms previous MIS methods in complex tasks

Abstract

Marginalized importance sampling (MIS), which measures the density ratio between the state-action occupancy of a target policy and that of a sampling distribution, is a promising approach for off-policy evaluation. However, current state-of-the-art MIS methods rely on complex optimization tricks and succeed mostly on simple toy problems. We bridge the gap between MIS and deep reinforcement learning by observing that the density ratio can be computed from the successor representation of the target policy. The successor representation can be trained through deep reinforcement learning methodology and decouples the reward optimization from the dynamics of the environment, making the resulting algorithm stable and applicable to high-dimensional domains. We evaluate the empirical performance of our approach on a variety of challenging Atari and MuJoCo environments.