Replay-Guided Adversarial Environment Design

TL;DR

This paper introduces a new perspective on environment design in reinforcement learning, unifying existing methods under a theoretical framework, and proposes a novel algorithm, PLR⊥, that improves out-of-distribution transfer performance.

Contribution

It formalizes Prioritized Level Replay as a form of Unsupervised Environment Design, develops a dual curriculum framework, and introduces PLR⊥, which enhances convergence and transfer in RL.

Findings

PLR⊥ outperforms existing methods on transfer tasks.

Theoretical guarantees are established for a class of UED methods.

Stopping policy updates on uncurated levels improves convergence.

Abstract

Deep reinforcement learning (RL) agents may successfully generalize to new settings if trained on an appropriately diverse set of environment and task configurations. Unsupervised Environment Design (UED) is a promising self-supervised RL paradigm, wherein the free parameters of an underspecified environment are automatically adapted during training to the agent's capabilities, leading to the emergence of diverse training environments. Here, we cast Prioritized Level Replay (PLR), an empirically successful but theoretically unmotivated method that selectively samples randomly-generated training levels, as UED. We argue that by curating completely random levels, PLR, too, can generate novel and complex levels for effective training. This insight reveals a natural class of UED methods we call Dual Curriculum Design (DCD). Crucially, DCD includes both PLR and a popular UED algorithm, PAIRED, as special cases and inherits similar theoretical guarantees. This connection allows us to develop novel theory for PLR, providing a version with a robustness guarantee at Nash equilibria. Furthermore, our theory suggests a highly counterintuitive improvement to PLR: by stopping the agent from updating its policy on uncurated levels (training on less data), we can improve the convergence to Nash equilibria. Indeed, our experiments confirm that our new method, PLR$^{\perp}$, obtains better results on a suite of out-of-distribution, zero-shot transfer tasks, in addition to demonstrating that PLR$^{\perp}$ improves the performance of PAIRED, from which it inherited its theoretical framework.