← 研踪 PaperTrailarXiv:2605.07537 · v2
Multi-Environment POMDPs with Finite-Horizon Objectives
首次提交 2026-05-08 18:14
版本更新 2026-11-13 18:01
雷达观测 2026-11-16 11:42
摘要
Partially Observable Markov Decision Processes (POMDPs) are systems in which one agent interacts with a stochastic environment, and receives only partial information about the current state. In a multi-environment POMDP (MEPOMDP), the initial state is unknown, and assumed to be adversarially chosen. In this work we focus on computing the optimal value and policy in MEPOMDPs with finite-horizon objectives. That problem is known to be PSPACE-complete in POMDPs. The extension is nontrivial: MEPOMDPs require reasoning about beliefs and payoffs across environments, and deterministic policies are no longer enough for optimal payoff. While the lower bound transfers immediately, the upper bound requires new techniques: we prove that mixtures of at most deterministic policies suffice for environments and show how to explore them without storing exponential-size policy descriptions. Our main results are as follows: (1) we establish that it is also PSPACE-complete in the more general setting of MEPOMDPs; (2) we present a practical algorithm and evaluate it on classical benchmarks, significantly outperforming the only previously known algorithm.
作者备注:33 pages, accepted at NeurIPS 2026 (Main Track)
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-13 18:01 · 最新观测版本
Multi-Environment POMDPs with Finite-Horizon Objectives
Partially Observable Markov Decision Processes (POMDPs) are systems in which one agent interacts with a stochastic environment, and receives only partial information about the current state. In a multi-environment POMDP (MEPOMDP), the initial state is unknown, and assumed to be adversarially chosen. In this work we focus on computing the optimal value and policy in MEPOMDPs with finite-horizon objectives. That problem is known to be PSPACE-complete in POMDPs. The extension is nontrivial: MEPOMDPs require reasoning about beliefs and payoffs across environments, and deterministic policies are no longer enough for optimal payoff. While the lower bound transfers immediately, the upper bound requires new techniques: we prove that mixtures of at most deterministic policies suffice for environments and show how to explore them without storing exponential-size policy descriptions. Our main results are as follows: (1) we establish that it is also PSPACE-complete in the more general setting of MEPOMDPs; (2) we present a practical algorithm and evaluate it on classical benchmarks, significantly outperforming the only previously known algorithm.
查看此版本解读 →查看此版本原文 ↗