Provably Safe Reinforcement Learning for Stochastic Reach-Avoid Problems with Entropy Regularization
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908774068912128 |
|---|---|
| author | Mazumdar, Abhijit Wisniewski, Rafal Bujorianu, Manuela L. |
| author_facet | Mazumdar, Abhijit Wisniewski, Rafal Bujorianu, Manuela L. |
| contents | We consider the problem of learning the optimal policy for Markov decision processes with safety constraints. We formulate the problem in a reach-avoid setup. Our goal is to design online reinforcement learning algorithms that ensure safety constraints with arbitrarily high probability during the learning phase. To this end, we first propose an algorithm based on the optimism in the face of uncertainty (OFU) principle. Based on the first algorithm, we propose our main algorithm, which utilizes entropy regularization. We investigate the finite-sample analysis of both algorithms and derive their regret bounds. We demonstrate that the inclusion of entropy regularization improves the regret and drastically controls the episode-to-episode variability that is inherent in OFU-based safe RL algorithms. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_08646 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Provably Safe Reinforcement Learning for Stochastic Reach-Avoid Problems with Entropy Regularization Mazumdar, Abhijit Wisniewski, Rafal Bujorianu, Manuela L. Machine Learning We consider the problem of learning the optimal policy for Markov decision processes with safety constraints. We formulate the problem in a reach-avoid setup. Our goal is to design online reinforcement learning algorithms that ensure safety constraints with arbitrarily high probability during the learning phase. To this end, we first propose an algorithm based on the optimism in the face of uncertainty (OFU) principle. Based on the first algorithm, we propose our main algorithm, which utilizes entropy regularization. We investigate the finite-sample analysis of both algorithms and derive their regret bounds. We demonstrate that the inclusion of entropy regularization improves the regret and drastically controls the episode-to-episode variability that is inherent in OFU-based safe RL algorithms. |
| title | Provably Safe Reinforcement Learning for Stochastic Reach-Avoid Problems with Entropy Regularization |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2601.08646 |