研究人员开发了一种新的策略外评估（OPE）方法，用于处理强化学习中奖励缺失非随机（MNAR）的情况。该方法利用未来状态作为影子变量来识别完整数据的条件平均奖励，从而解决选择偏差问题。提出的估计器受Fitted-Q-Evaluation的启发，允许目标策略纳入过去的缺失指标，并在模拟数据和MIMIC-III败血症数据集的实验中表现出强大的性能。 AI
影响 提高了强化学习模型在数据不完整现实场景中的可靠性。
排序理由 关于机器学习新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2606.20206v1 Announce Type: new Abstract: In offline Reinforcement Learning, immediate rewards in logged batch data are often unobserved due to sparse or irregular record-keeping, or censored beyond certain reward values. This issue arises in practical settings, including h…
In offline Reinforcement Learning, immediate rewards in logged batch data are often unobserved due to sparse or irregular record-keeping, or censored beyond certain reward values. This issue arises in practical settings, including health care and marketing. We investigate off-pol…