Matches in SemOpenAlex for { <https://semopenalex.org/work/W4285606945> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W4285606945 abstract "Though deep reinforcement learning (DRL) has obtained substantial success, it may encounter catastrophic failures due to the intrinsic uncertainty of both transition and observation. Most of the existing methods for safe reinforcement learning can only handle transition disturbance or observation disturbance since these two kinds of disturbance affect different parts of the agent; besides, the popular worst-case return may lead to overly pessimistic policies. To address these issues, we first theoretically prove that the performance degradation under transition disturbance and observation disturbance depends on a novel metric of Value Function Range (VFR), which corresponds to the gap in the value function between the best state and the worst state. Based on the analysis, we adopt conditional value-at-risk (CVaR) as an assessment of risk and propose a novel reinforcement learning algorithm of CVaR-Proximal-Policy-Optimization (CPPO) which formalizes the risk-sensitive constrained optimization problem by keeping its CVaR under a given threshold. Experimental results show that CPPO achieves a higher cumulative reward and is more robust against both observation and transition disturbances on a series of continuous control tasks in MuJoCo." @default.
- W4285606945 created "2022-07-16" @default.
- W4285606945 creator A5032457469 @default.
- W4285606945 creator A5060291357 @default.
- W4285606945 creator A5067301020 @default.
- W4285606945 date "2022-07-01" @default.
- W4285606945 modified "2023-10-15" @default.
- W4285606945 title "Towards Safe Reinforcement Learning via Constraining Conditional Value-at-Risk" @default.
- W4285606945 doi "https://doi.org/10.24963/ijcai.2022/510" @default.
- W4285606945 hasPublicationYear "2022" @default.
- W4285606945 type Work @default.
- W4285606945 citedByCount "1" @default.
- W4285606945 countsByYear W42856069452023 @default.
- W4285606945 crossrefType "proceedings-article" @default.
- W4285606945 hasAuthorship W4285606945A5032457469 @default.
- W4285606945 hasAuthorship W4285606945A5060291357 @default.
- W4285606945 hasAuthorship W4285606945A5067301020 @default.
- W4285606945 hasBestOaLocation W42856069451 @default.
- W4285606945 hasConcept C111472728 @default.
- W4285606945 hasConcept C126255220 @default.
- W4285606945 hasConcept C127413603 @default.
- W4285606945 hasConcept C138885662 @default.
- W4285606945 hasConcept C14036430 @default.
- W4285606945 hasConcept C14646407 @default.
- W4285606945 hasConcept C146978453 @default.
- W4285606945 hasConcept C151730666 @default.
- W4285606945 hasConcept C154945302 @default.
- W4285606945 hasConcept C162324750 @default.
- W4285606945 hasConcept C176217482 @default.
- W4285606945 hasConcept C187736073 @default.
- W4285606945 hasConcept C204323151 @default.
- W4285606945 hasConcept C21547014 @default.
- W4285606945 hasConcept C2775924081 @default.
- W4285606945 hasConcept C2777601987 @default.
- W4285606945 hasConcept C2779922397 @default.
- W4285606945 hasConcept C32896092 @default.
- W4285606945 hasConcept C33923547 @default.
- W4285606945 hasConcept C41008148 @default.
- W4285606945 hasConcept C47446073 @default.
- W4285606945 hasConcept C5496284 @default.
- W4285606945 hasConcept C66938386 @default.
- W4285606945 hasConcept C67203356 @default.
- W4285606945 hasConcept C78458016 @default.
- W4285606945 hasConcept C86803240 @default.
- W4285606945 hasConcept C97541855 @default.
- W4285606945 hasConcept C9992130 @default.
- W4285606945 hasConceptScore W4285606945C111472728 @default.
- W4285606945 hasConceptScore W4285606945C126255220 @default.
- W4285606945 hasConceptScore W4285606945C127413603 @default.
- W4285606945 hasConceptScore W4285606945C138885662 @default.
- W4285606945 hasConceptScore W4285606945C14036430 @default.
- W4285606945 hasConceptScore W4285606945C14646407 @default.
- W4285606945 hasConceptScore W4285606945C146978453 @default.
- W4285606945 hasConceptScore W4285606945C151730666 @default.
- W4285606945 hasConceptScore W4285606945C154945302 @default.
- W4285606945 hasConceptScore W4285606945C162324750 @default.
- W4285606945 hasConceptScore W4285606945C176217482 @default.
- W4285606945 hasConceptScore W4285606945C187736073 @default.
- W4285606945 hasConceptScore W4285606945C204323151 @default.
- W4285606945 hasConceptScore W4285606945C21547014 @default.
- W4285606945 hasConceptScore W4285606945C2775924081 @default.
- W4285606945 hasConceptScore W4285606945C2777601987 @default.
- W4285606945 hasConceptScore W4285606945C2779922397 @default.
- W4285606945 hasConceptScore W4285606945C32896092 @default.
- W4285606945 hasConceptScore W4285606945C33923547 @default.
- W4285606945 hasConceptScore W4285606945C41008148 @default.
- W4285606945 hasConceptScore W4285606945C47446073 @default.
- W4285606945 hasConceptScore W4285606945C5496284 @default.
- W4285606945 hasConceptScore W4285606945C66938386 @default.
- W4285606945 hasConceptScore W4285606945C67203356 @default.
- W4285606945 hasConceptScore W4285606945C78458016 @default.
- W4285606945 hasConceptScore W4285606945C86803240 @default.
- W4285606945 hasConceptScore W4285606945C97541855 @default.
- W4285606945 hasConceptScore W4285606945C9992130 @default.
- W4285606945 hasLocation W42856069451 @default.
- W4285606945 hasLocation W42856069452 @default.
- W4285606945 hasOpenAccess W4285606945 @default.
- W4285606945 hasPrimaryLocation W42856069451 @default.
- W4285606945 hasRelatedWork W1997088969 @default.
- W4285606945 hasRelatedWork W2002203564 @default.
- W4285606945 hasRelatedWork W3098129903 @default.
- W4285606945 hasRelatedWork W3126217842 @default.
- W4285606945 hasRelatedWork W3214094365 @default.
- W4285606945 hasRelatedWork W4226345898 @default.
- W4285606945 hasRelatedWork W4285606945 @default.
- W4285606945 hasRelatedWork W4287331027 @default.
- W4285606945 hasRelatedWork W4309134502 @default.
- W4285606945 hasRelatedWork W4362656159 @default.
- W4285606945 isParatext "false" @default.
- W4285606945 isRetracted "false" @default.
- W4285606945 workType "article" @default.