Curious Replay：用于基于模型的适应的好奇回放
Curious Replay for Model-based Adaptation
AI总结：
针对基于模型的强化学习智能体在环境变化时适应能力不足的问题，提出 Curious Replay 方法，利用好奇心优先级信号改进世界模型训练，在 Crafter 基准上取得 19.4 的平均分超越现有最佳性能。
摘要：
智能体必须能够在环境变化时快速适应。我们发现，现有的基于模型的强化学习智能体无法很好地做到这一点，部分原因在于它们如何使用过往经验来训练其世界模型。在此，我们提出 Curious Replay——一种通过使用基于好奇心的优先级信号，为基于模型的智能体量身定制的优先经验回放形式。使用 Curious Replay 的智能体在受动物行为启发的探索范式以及 Crafter 基准上展现出更好的性能。结合 Curious Replay 的 DreamerV3 在 Crafter 上超越了现有最佳性能，取得了 19.4 的平均分，大幅优于使用均匀回放的 DreamerV3 之前 14.5 的高分，同时在 Deepmind Control Suite 上保持相似的性能。Curious Replay 的代码可在 https://github.com/AutonomousAgentsLab/curiousreplay 获取。
英文摘要：
Agents must be able to adapt quickly as an environment changes. We find that existing model-based reinforcement learning agents are unable to do this well, in part because of how they use past experiences to train their world model. Here, we present Curious Replay -- a form of prioritized experience replay tailored to model-based agents through use of a curiosity-based priority signal. Agents using Curious Replay exhibit improved performance in an exploration paradigm inspired by animal behavior and on the Crafter benchmark. DreamerV3 with Curious Replay surpasses state-of-the-art performance on Crafter, achieving a mean score of 19.4 that substantially improves on the previous high score of 14.5 by DreamerV3 with uniform replay, while also maintaining similar performance on the Deepmind Control Suite. Code for Curious Replay is available at https://github.com/AutonomousAgentsLab/curiousreplay
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
