Inferring Reward Machines and Transition Machines from Partially Observable Markov Decision Processes
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909720472715264 |
|---|---|
| author | Wu, Yuly Liu, Jiamou Zhang, Libo |
| author_facet | Wu, Yuly Liu, Jiamou Zhang, Libo |
| contents | Partially Observable Markov Decision Processes (POMDPs) are fundamental to many real-world applications. Although reinforcement learning (RL) has shown success in fully observable domains, learning policies from traces in partially observable environments remains challenging due to non-Markovian observations. Inferring an automaton to handle the non-Markovianity is a proven effective approach, but faces two limitations: 1) existing automaton representations focus only on reward-based non-Markovianity, leading to unnatural problem formulations; 2) inference algorithms face enormous computational costs. For the first limitation, we introduce Transition Machines (TMs) to complement existing Reward Machines (RMs). To develop a unified inference algorithm for both automata types, we propose the Dual Behavior Mealy Machine (DBMM) that subsumes both TMs and RMs. We then introduce DB-RPNI, a passive automata learning algorithm that efficiently infers DBMMs while avoiding the costly reductions required by prior work. We further develop optimization techniques and identify sufficient conditions for inferring the minimal correct automata. Experimentally, our inference method achieves speedups of up to three orders of magnitude over SOTA baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_01947 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Inferring Reward Machines and Transition Machines from Partially Observable Markov Decision Processes Wu, Yuly Liu, Jiamou Zhang, Libo Machine Learning Artificial Intelligence I.2 Partially Observable Markov Decision Processes (POMDPs) are fundamental to many real-world applications. Although reinforcement learning (RL) has shown success in fully observable domains, learning policies from traces in partially observable environments remains challenging due to non-Markovian observations. Inferring an automaton to handle the non-Markovianity is a proven effective approach, but faces two limitations: 1) existing automaton representations focus only on reward-based non-Markovianity, leading to unnatural problem formulations; 2) inference algorithms face enormous computational costs. For the first limitation, we introduce Transition Machines (TMs) to complement existing Reward Machines (RMs). To develop a unified inference algorithm for both automata types, we propose the Dual Behavior Mealy Machine (DBMM) that subsumes both TMs and RMs. We then introduce DB-RPNI, a passive automata learning algorithm that efficiently infers DBMMs while avoiding the costly reductions required by prior work. We further develop optimization techniques and identify sufficient conditions for inferring the minimal correct automata. Experimentally, our inference method achieves speedups of up to three orders of magnitude over SOTA baselines. |
| title | Inferring Reward Machines and Transition Machines from Partially Observable Markov Decision Processes |
| topic | Machine Learning Artificial Intelligence I.2 |
| url | https://arxiv.org/abs/2508.01947 |