Salvato in:
Dettagli Bibliografici
Autori principali: Alyahya, Abdulaziz, Siyabi, Abdallah Al, Ernst, Markus R., Yang, Luke, Kuhlmann, Levin, Kowadlo, Gideon
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2603.11395
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916019579125760
author Alyahya, Abdulaziz
Siyabi, Abdallah Al
Ernst, Markus R.
Yang, Luke
Kuhlmann, Levin
Kowadlo, Gideon
author_facet Alyahya, Abdulaziz
Siyabi, Abdallah Al
Ernst, Markus R.
Yang, Luke
Kuhlmann, Levin
Kowadlo, Gideon
contents Continual reinforcement learning challenges agents to acquire new skills while retaining previously learned ones with the goal of improving performance in both past and future tasks. Most existing approaches rely on model-free methods with replay buffers to mitigate catastrophic forgetting; however, these solutions often face significant scalability challenges due to large memory demands. Drawing inspiration from neuroscience, where the brain replays experiences to a predictive World Model rather than directly to the policy, we present ARROW (Augmented Replay for RObust World models), a model-based continual RL algorithm that extends DreamerV3 with a memory-efficient, distribution-matching replay buffer. Unlike standard fixed-size FIFO buffers, ARROW maintains two complementary buffers: a short-term buffer for recent experiences and a long-term buffer that preserves task diversity through intelligent sampling. We evaluate ARROW on two challenging continual RL settings: Tasks without shared structure (Atari), and tasks with shared structure, where knowledge transfer is possible (Procgen CoinRun variants). Compared to model-free and model-based baselines with replay buffers of the same-size, ARROW demonstrates substantially less forgetting on tasks without shared structure, while maintaining comparable forward transfer. Our findings highlight the potential of model-based RL and bio-inspired approaches for continual reinforcement learning, warranting further research.
format Preprint
id arxiv_https___arxiv_org_abs_2603_11395
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ARROW: Augmented Replay for RObust World models
Alyahya, Abdulaziz
Siyabi, Abdallah Al
Ernst, Markus R.
Yang, Luke
Kuhlmann, Levin
Kowadlo, Gideon
Machine Learning
Artificial Intelligence
Continual reinforcement learning challenges agents to acquire new skills while retaining previously learned ones with the goal of improving performance in both past and future tasks. Most existing approaches rely on model-free methods with replay buffers to mitigate catastrophic forgetting; however, these solutions often face significant scalability challenges due to large memory demands. Drawing inspiration from neuroscience, where the brain replays experiences to a predictive World Model rather than directly to the policy, we present ARROW (Augmented Replay for RObust World models), a model-based continual RL algorithm that extends DreamerV3 with a memory-efficient, distribution-matching replay buffer. Unlike standard fixed-size FIFO buffers, ARROW maintains two complementary buffers: a short-term buffer for recent experiences and a long-term buffer that preserves task diversity through intelligent sampling. We evaluate ARROW on two challenging continual RL settings: Tasks without shared structure (Atari), and tasks with shared structure, where knowledge transfer is possible (Procgen CoinRun variants). Compared to model-free and model-based baselines with replay buffers of the same-size, ARROW demonstrates substantially less forgetting on tasks without shared structure, while maintaining comparable forward transfer. Our findings highlight the potential of model-based RL and bio-inspired approaches for continual reinforcement learning, warranting further research.
title ARROW: Augmented Replay for RObust World models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2603.11395