Matches in SemOpenAlex for { <https://semopenalex.org/work/W3100407858> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W3100407858 endingPage "6183" @default.
- W3100407858 startingPage "6172" @default.
- W3100407858 abstract "We present Revel, a partially neural reinforcement learning (RL) framework for provably safe exploration in continuous state and action spaces. A key challenge for provably safe deep RL is that repeatedly verifying neural networks within a learning loop is computationally infeasible. We address this challenge using two policy classes: a general, neurosymbolic class with approximate gradients and a more restricted class of symbolic policies that allows efficient verification. Our learning algorithm is a mirror descent over policies: in each iteration, it safely lifts a symbolic policy into the neurosymbolic space, performs safe gradient updates to the resulting policy, and projects the updated policy into the safe symbolic subset, all without requiring explicit verification of neural networks. Our empirical results show that Revel enforces safe exploration in many scenarios in which Constrained Policy Optimization does not, and that it can discover policies that outperform those learned through prior approaches to verified exploration." @default.
- W3100407858 created "2020-11-23" @default.
- W3100407858 creator A5006424908 @default.
- W3100407858 creator A5026149314 @default.
- W3100407858 creator A5057341982 @default.
- W3100407858 creator A5067492876 @default.
- W3100407858 date "2020-01-01" @default.
- W3100407858 modified "2023-10-18" @default.
- W3100407858 title "Neurosymbolic Reinforcement Learning with Formally Verified Exploration" @default.
- W3100407858 hasPublicationYear "2020" @default.
- W3100407858 type Work @default.
- W3100407858 sameAs 3100407858 @default.
- W3100407858 citedByCount "5" @default.
- W3100407858 countsByYear W31004078582021 @default.
- W3100407858 crossrefType "proceedings-article" @default.
- W3100407858 hasAuthorship W3100407858A5006424908 @default.
- W3100407858 hasAuthorship W3100407858A5026149314 @default.
- W3100407858 hasAuthorship W3100407858A5057341982 @default.
- W3100407858 hasAuthorship W3100407858A5067492876 @default.
- W3100407858 hasConcept C105795698 @default.
- W3100407858 hasConcept C11413529 @default.
- W3100407858 hasConcept C119857082 @default.
- W3100407858 hasConcept C147168706 @default.
- W3100407858 hasConcept C153258448 @default.
- W3100407858 hasConcept C154945302 @default.
- W3100407858 hasConcept C26517878 @default.
- W3100407858 hasConcept C2777212361 @default.
- W3100407858 hasConcept C2779436431 @default.
- W3100407858 hasConcept C33923547 @default.
- W3100407858 hasConcept C38652104 @default.
- W3100407858 hasConcept C41008148 @default.
- W3100407858 hasConcept C48103436 @default.
- W3100407858 hasConcept C50644808 @default.
- W3100407858 hasConcept C72434380 @default.
- W3100407858 hasConcept C80444323 @default.
- W3100407858 hasConcept C97541855 @default.
- W3100407858 hasConceptScore W3100407858C105795698 @default.
- W3100407858 hasConceptScore W3100407858C11413529 @default.
- W3100407858 hasConceptScore W3100407858C119857082 @default.
- W3100407858 hasConceptScore W3100407858C147168706 @default.
- W3100407858 hasConceptScore W3100407858C153258448 @default.
- W3100407858 hasConceptScore W3100407858C154945302 @default.
- W3100407858 hasConceptScore W3100407858C26517878 @default.
- W3100407858 hasConceptScore W3100407858C2777212361 @default.
- W3100407858 hasConceptScore W3100407858C2779436431 @default.
- W3100407858 hasConceptScore W3100407858C33923547 @default.
- W3100407858 hasConceptScore W3100407858C38652104 @default.
- W3100407858 hasConceptScore W3100407858C41008148 @default.
- W3100407858 hasConceptScore W3100407858C48103436 @default.
- W3100407858 hasConceptScore W3100407858C50644808 @default.
- W3100407858 hasConceptScore W3100407858C72434380 @default.
- W3100407858 hasConceptScore W3100407858C80444323 @default.
- W3100407858 hasConceptScore W3100407858C97541855 @default.
- W3100407858 hasLocation W31004078581 @default.
- W3100407858 hasOpenAccess W3100407858 @default.
- W3100407858 hasPrimaryLocation W31004078581 @default.
- W3100407858 hasRelatedWork W2399554456 @default.
- W3100407858 hasRelatedWork W273478400 @default.
- W3100407858 hasRelatedWork W2749952662 @default.
- W3100407858 hasRelatedWork W2791704483 @default.
- W3100407858 hasRelatedWork W2809801155 @default.
- W3100407858 hasRelatedWork W2951456741 @default.
- W3100407858 hasRelatedWork W2951543502 @default.
- W3100407858 hasRelatedWork W3007369745 @default.
- W3100407858 hasRelatedWork W3035521307 @default.
- W3100407858 hasRelatedWork W3040161731 @default.
- W3100407858 hasRelatedWork W3045280543 @default.
- W3100407858 hasRelatedWork W3088154658 @default.
- W3100407858 hasRelatedWork W3133143524 @default.
- W3100407858 hasRelatedWork W3136683507 @default.
- W3100407858 hasRelatedWork W3159199672 @default.
- W3100407858 hasRelatedWork W3192708540 @default.
- W3100407858 hasRelatedWork W3198607174 @default.
- W3100407858 hasRelatedWork W3204843673 @default.
- W3100407858 hasRelatedWork W3213845467 @default.
- W3100407858 hasRelatedWork W76312321 @default.
- W3100407858 hasVolume "33" @default.
- W3100407858 isParatext "false" @default.
- W3100407858 isRetracted "false" @default.
- W3100407858 magId "3100407858" @default.
- W3100407858 workType "article" @default.