Counterexample-Guided Repair of Reinforcement Learning Systems Using Safety Critics
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929357475282944 |
|---|---|
| author | Boetius, David Leue, Stefan |
| author_facet | Boetius, David Leue, Stefan |
| contents | Naively trained Deep Reinforcement Learning agents may fail to satisfy vital safety constraints. To avoid costly retraining, we may desire to repair a previously trained reinforcement learning agent to obviate unsafe behaviour. We devise a counterexample-guided repair algorithm for repairing reinforcement learning systems leveraging safety critics. The algorithm jointly repairs a reinforcement learning agent and a safety critic using gradient-based constrained optimisation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_15430 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Counterexample-Guided Repair of Reinforcement Learning Systems Using Safety Critics Boetius, David Leue, Stefan Machine Learning Logic in Computer Science Naively trained Deep Reinforcement Learning agents may fail to satisfy vital safety constraints. To avoid costly retraining, we may desire to repair a previously trained reinforcement learning agent to obviate unsafe behaviour. We devise a counterexample-guided repair algorithm for repairing reinforcement learning systems leveraging safety critics. The algorithm jointly repairs a reinforcement learning agent and a safety critic using gradient-based constrained optimisation. |
| title | Counterexample-Guided Repair of Reinforcement Learning Systems Using Safety Critics |
| topic | Machine Learning Logic in Computer Science |
| url | https://arxiv.org/abs/2405.15430 |