Metraipolis › archive › dc_f677018fe7cc

Adaptive Probabilistic Shielding by Learning MDPs for Safe Reinforcement Learning — arXiv:2608.19836

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.19836
License (as archived)
CC0-1.0
Archived
2026-10-18T11:48:17.397Z
SHA-256 of the archived text
f5c0224e0749a93fb03dc5c28f63706f6138e686ec85efa9d3530c3160b8a464
Stable URL of this page
https://metraipolis.com/documento/dc_f677018fe7cc

How to cite

Adaptive Probabilistic Shielding by Learning MDPs for Safe Reinforcement Learning — arXiv:2608.19836 — arXiv (en), https://arxiv.org/abs/2608.19836 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f677018fe7cc (Metraipolis archive dc_f677018fe7cc, sha256 f5c0224e0749a93fb03dc5c28f63706f6138e686ec85efa9d3530c3160b8a464)
BibTeX
@misc{dcf677018fe7cc,
  title = {Adaptive Probabilistic Shielding by Learning MDPs for Safe Reinforcement Learning — arXiv:2608.19836},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.19836},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f677018fe7cc; sha256 f5c0224e0749a93fb03dc5c28f63706f6138e686ec85efa9d3530c3160b8a464},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.19836. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Probabilistic shielding is a technique for safe reinforcement learning (RL). Typically, a static observer -- called the shield -- constrains the learning agent's actions to those for which acting safely remains feasible. Traditionally, the shield is computed from the transition probabilities of the underlying Markov decision process (MDP). Thus, this technique is not applicable when the MDP model is not given a priori, which, unfortunately, is the case in typical RL applications. In this paper, we study the problem of computing a shield in the setting where the transition graph of the MDP is known, but the transition probabilities are unknown. Our approach integrates probabilistic shielding with online model learning: as the RL agent explores the environment, we estimate the transition probabilities. From this estimate, we compute a shield. While the shield may be conservative initially, it adapts as the model estimate becomes more precise. Thus, the shield improves in tandem with the RL agent. This paradigm of adaptive probabilistic shielding raises a number of challenges, such as when to recompute the shield and how to balance between exploration and safety during learning. We empirically evaluate multiple variants of this paradigm across several environments.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-21 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO).
Pagina del lavoro: https://arxiv.org/abs/2608.19836 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T11:47:43.320Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).