arXiv · AI· Durgam Latha, Dion Reji, Dorde Zikelic, Shankaranarayanan Krishna·· 12 小时前AI 评分22
R(PO)MDP 奖励目标与 (PO)游戏 的保持律等价性分析
Reward Objectives for Robust (PO)MDPs: A Law-Preserving Equivalence with (Partially Observable) Games
AI 导读
研究提出统一框架，通过将 (s,a)-rectangular RPOMDP 归约 partially observable stochastic games (POSGs) 解决鲁棒决策问题。
来源：arXiv · AI · arxiv.org