Bootstrap Latent-Predictive Representations for Multitask Reinforcement Learning

TL;DR

This paper introduces PBL, a self-supervised representation learning method for multitask deep reinforcement learning that predicts future latent embeddings to improve agent performance across various environments.

Contribution

The paper presents PBL, a novel latent-predictive representation learning algorithm that enhances multitask RL by capturing environment dynamics through bootstrap predictions in latent space.

Findings

PBL outperforms state-of-the-art RL agents on DMLab-30 and Atari-57 benchmarks.

PBL effectively handles multimodal observations including images and language.

The method improves learning efficiency and task performance in complex environments.

Abstract

Learning a good representation is an essential component for deep reinforcement learning (RL). Representation learning is especially important in multitask and partially observable settings where building a representation of the unknown environment is crucial to solve the tasks. Here we introduce Prediction of Bootstrap Latents (PBL), a simple and flexible self-supervised representation learning algorithm for multitask deep RL. PBL builds on multistep predictive representations of future observations, and focuses on capturing structured information about environment dynamics. Specifically, PBL trains its representation by predicting latent embeddings of future observations. These latent embeddings are themselves trained to be predictive of the aforementioned representations. These predictions form a bootstrapping effect, allowing the agent to learn more about the key aspects of the environment dynamics. In addition, by defining prediction tasks completely in latent space, PBL provides the flexibility of using multimodal observations involving pixel images, language instructions, rewards and more. We show in our experiments that PBL delivers across-the-board improved performance over state of the art deep RL agents in the DMLab-30 and Atari-57 multitask setting.