The Effectiveness of World Models for Continual Reinforcement Learning

Autor:	Kessler, Samuel, Ostaszewski, Mateusz, Bortkiewicz, Michał, Żarski, Mateusz, Wołczyk, Maciej, Parker-Holder, Jack, Roberts, Stephen J., Miłoś, Piotr
Rok vydání:	2022
Předmět:	Computer Science - Machine Learning Computer Science - Artificial Intelligence
Druh dokumentu:	Working Paper
Popis:	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. Comment: Accepted at CoLLAs 2023, 21 pages, 15 figures
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2211.15944 Zobrazit plný text záznamu View this record from Arxiv