Learning Successor Features the Simple Way

TL;DR

This paper introduces a simple, efficient method for learning Successor Features directly from pixel data in reinforcement learning, avoiding complex losses and outperforming existing techniques in various environments.

Contribution

A novel, straightforward approach combining TD and reward prediction losses to learn SFs from pixels without pretraining or complex procedures.

Findings

Matches or outperforms existing SF learning methods

Effective in 2D, 3D, and Mujoco environments

Achieves higher performance faster than previous approaches

Abstract

In Deep Reinforcement Learning (RL), it is a challenge to learn representations that do not exhibit catastrophic forgetting or interference in non-stationary environments. Successor Features (SFs) offer a potential solution to this challenge. However, canonical techniques for learning SFs from pixel-level observations often lead to representation collapse, wherein representations degenerate and fail to capture meaningful variations in the data. More recent methods for learning SFs can avoid representation collapse, but they often involve complex losses and multiple learning phases, reducing their efficiency. We introduce a novel, simple method for learning SFs directly from pixels. Our approach uses a combination of a Temporal-difference (TD) loss and a reward prediction loss, which together capture the basic mathematical definition of SFs. We show that our approach matches or outperforms existing SF learning techniques in both 2D (Minigrid), 3D (Miniworld) mazes and Mujoco, for both single and continual learning scenarios. As well, our technique is efficient, and can reach higher levels of performance in less time than other approaches. Our work provides a new, streamlined technique for learning SFs directly from pixel observations, with no pretraining required.