研究人员引入了拟合占用率评估（FORE），一种用于离线强化学习中估计占用率的新颖方法。该技术通过伴随贝尔曼递归来表征折扣占用率，并在每次迭代中解决密度比目标。FORE的关键创新在于其简化的近似条件，仅需要折扣占用率本身的实现性，而不是像贝尔曼完备性那样更复杂的条件。这种方法能够直接进行价值估计和双重鲁棒估计，为离线策略评估提供了更鲁棒的方法。 AI
影响 通过放宽严格的数学条件，为离线强化学习评估引入了更鲁棒的方法。
排序理由 该集群包含一篇详细介绍强化学习新方法的学术论文。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
Occupancy ratios correct distribution shift in offline reinforcement learning and are central to off-policy evaluation. Existing primal-dual and minimax methods typically estimate these ratios by enforcing occupancy-balance moments over a critic class. We propose fitted occupancy…
arXiv:2607.05375v1 Announce Type: new Abstract: Occupancy ratios correct distribution shift in offline reinforcement learning and are central to off-policy evaluation. Existing primal-dual and minimax methods typically estimate these ratios by enforcing occupancy-balance moments …
Occupancy ratios correct distribution shift in offline reinforcement learning and are central to off-policy evaluation. Existing primal-dual and minimax methods typically estimate these ratios by enforcing occupancy-balance moments over a critic class. We propose fitted occupancy…