Tighter Value-Function Approximations for POMDPs
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929707127144448 |
|---|---|
| author | Krale, Merlijn Koops, Wietze Junges, Sebastian Simão, Thiago D. Jansen, Nils |
| author_facet | Krale, Merlijn Koops, Wietze Junges, Sebastian Simão, Thiago D. Jansen, Nils |
| contents | Solving partially observable Markov decision processes (POMDPs) typically requires reasoning about the values of exponentially many state beliefs. Towards practical performance, state-of-the-art solvers use value bounds to guide this reasoning. However, sound upper value bounds are often computationally expensive to compute, and there is a tradeoff between the tightness of such bounds and their computational cost. This paper introduces new and provably tighter upper value bounds than the commonly used fast informed bound. Our empirical evaluation shows that, despite their additional computational overhead, the new upper bounds accelerate state-of-the-art POMDP solvers on a wide range of benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_06523 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Tighter Value-Function Approximations for POMDPs Krale, Merlijn Koops, Wietze Junges, Sebastian Simão, Thiago D. Jansen, Nils Artificial Intelligence Solving partially observable Markov decision processes (POMDPs) typically requires reasoning about the values of exponentially many state beliefs. Towards practical performance, state-of-the-art solvers use value bounds to guide this reasoning. However, sound upper value bounds are often computationally expensive to compute, and there is a tradeoff between the tightness of such bounds and their computational cost. This paper introduces new and provably tighter upper value bounds than the commonly used fast informed bound. Our empirical evaluation shows that, despite their additional computational overhead, the new upper bounds accelerate state-of-the-art POMDP solvers on a wide range of benchmarks. |
| title | Tighter Value-Function Approximations for POMDPs |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2502.06523 |