Introducing Symmetries to Black Box Meta Reinforcement Learning

TL;DR

This paper investigates how incorporating symmetries into black-box meta reinforcement learning systems can enhance their ability to generalize across unseen environments, actions, and observations.

Contribution

It introduces a new black-box meta RL system that explicitly incorporates symmetries, improving generalization capabilities over traditional approaches.

Findings

Symmetries improve generalization to unseen environments.

The proposed system outperforms traditional black-box meta RL methods.

Incorporating symmetries enhances transferability across tasks.

Abstract

Meta reinforcement learning (RL) attempts to discover new RL algorithms automatically from environment interaction. In so-called black-box approaches, the policy and the learning algorithm are jointly represented by a single neural network. These methods are very flexible, but they tend to underperform in terms of generalisation to new, unseen environments. In this paper, we explore the role of symmetries in meta-generalisation. We show that a recent successful meta RL approach that meta-learns an objective for backpropagation-based learning exhibits certain symmetries (specifically the reuse of the learning rule, and invariance to input and output permutations) that are not present in typical black-box meta RL systems. We hypothesise that these symmetries can play an important role in meta-generalisation. Building off recent work in black-box supervised meta learning, we develop a black-box meta RL system that exhibits these same symmetries. We show through careful experimentation that incorporating these symmetries can lead to algorithms with a greater ability to generalise to unseen action & observation spaces, tasks, and environments.