The Effectiveness of World Models for Continual Reinforcement Learning

TL;DR

This paper demonstrates that world models can be effectively adapted for continual reinforcement learning, improving performance and transfer in changing environments through selective experience replay and new modeling strategies.

Contribution

It introduces Continual-Dreamer, a task-agnostic world model approach for continual RL that outperforms existing methods on benchmark tasks.

Findings

Continual-Dreamer is sample efficient.

It outperforms state-of-the-art continual RL methods.

Selective experience replay enhances performance and transfer.

Abstract

World models power some of the most efficient reinforcement learning algorithms. In this work, we showcase that they can be harnessed for continual learning - a situation when the agent faces changing environments. World models typically employ a replay buffer for training, which can be naturally extended to continual learning. We systematically study how different selective experience replay methods affect performance, forgetting, and transfer. We also provide recommendations regarding various modeling options for using world models. The best set of choices is called Continual-Dreamer, it is task-agnostic and utilizes the world model for continual exploration. Continual-Dreamer is sample efficient and outperforms state-of-the-art task-agnostic continual reinforcement learning methods on Minigrid and Minihack benchmarks.