Matches in SemOpenAlex for { <https://semopenalex.org/work/W3209383860> ?p ?o ?g. }
- W3209383860 abstract "Recently, a variety of new equivariant neural network model architectures have been proposed that generalize better over rotational and reflectional symmetries than standard models. These models are relevant to robotics because many robotics problems can be expressed in a rotationally symmetric way. This paper focuses on equivariance over a visual state space and a spatial action space -- the setting where the robot action space includes a subset of $rm{SE}(2)$. In this situation, we know a priori that rotations and translations in the state image should result in the same rotations and translations in the spatial action dimensions of the optimal policy. Therefore, we can use equivariant model architectures to make $Q$ learning more sample efficient. This paper identifies when the optimal $Q$ function is equivariant and proposes $Q$ network architectures for this setting. We show experimentally that this approach outperforms standard methods in a set of challenging manipulation problems." @default.
- W3209383860 created "2021-11-08" @default.
- W3209383860 creator A5020217201 @default.
- W3209383860 creator A5035797556 @default.
- W3209383860 creator A5046522371 @default.
- W3209383860 creator A5066935778 @default.
- W3209383860 date "2021-10-28" @default.
- W3209383860 modified "2023-09-25" @default.
- W3209383860 title "Equivariant $Q$ Learning in Spatial Action Spaces" @default.
- W3209383860 cites W1901129140 @default.
- W3209383860 cites W2021683594 @default.
- W3209383860 cites W2046033161 @default.
- W3209383860 cites W2112796928 @default.
- W3209383860 cites W2163605009 @default.
- W3209383860 cites W2279221249 @default.
- W3209383860 cites W2336763592 @default.
- W3209383860 cites W2414711238 @default.
- W3209383860 cites W2842511635 @default.
- W3209383860 cites W2899771611 @default.
- W3209383860 cites W2914944935 @default.
- W3209383860 cites W2962746398 @default.
- W3209383860 cites W2962793652 @default.
- W3209383860 cites W2962965579 @default.
- W3209383860 cites W2963033241 @default.
- W3209383860 cites W2963092423 @default.
- W3209383860 cites W2963376229 @default.
- W3209383860 cites W2963477884 @default.
- W3209383860 cites W2964121744 @default.
- W3209383860 cites W2970754912 @default.
- W3209383860 cites W2970756352 @default.
- W3209383860 cites W2979421998 @default.
- W3209383860 cites W3004003541 @default.
- W3209383860 cites W3016361525 @default.
- W3209383860 cites W3021708257 @default.
- W3209383860 cites W3023640063 @default.
- W3209383860 cites W3036642894 @default.
- W3209383860 cites W3037007134 @default.
- W3209383860 cites W3038745179 @default.
- W3209383860 cites W3091677803 @default.
- W3209383860 cites W3092364399 @default.
- W3209383860 cites W3098082936 @default.
- W3209383860 cites W3098143573 @default.
- W3209383860 cites W3102762742 @default.
- W3209383860 cites W3107359931 @default.
- W3209383860 cites W3111393852 @default.
- W3209383860 cites W3115293622 @default.
- W3209383860 cites W3119794164 @default.
- W3209383860 cites W3210940825 @default.
- W3209383860 hasPublicationYear "2021" @default.
- W3209383860 type Work @default.
- W3209383860 sameAs 3209383860 @default.
- W3209383860 citedByCount "0" @default.
- W3209383860 crossrefType "posted-content" @default.
- W3209383860 hasAuthorship W3209383860A5020217201 @default.
- W3209383860 hasAuthorship W3209383860A5035797556 @default.
- W3209383860 hasAuthorship W3209383860A5046522371 @default.
- W3209383860 hasAuthorship W3209383860A5066935778 @default.
- W3209383860 hasBestOaLocation W32093838601 @default.
- W3209383860 hasConcept C105795698 @default.
- W3209383860 hasConcept C111472728 @default.
- W3209383860 hasConcept C111919701 @default.
- W3209383860 hasConcept C11413529 @default.
- W3209383860 hasConcept C114614502 @default.
- W3209383860 hasConcept C121332964 @default.
- W3209383860 hasConcept C136197465 @default.
- W3209383860 hasConcept C138885662 @default.
- W3209383860 hasConcept C14036430 @default.
- W3209383860 hasConcept C154945302 @default.
- W3209383860 hasConcept C171036898 @default.
- W3209383860 hasConcept C177264268 @default.
- W3209383860 hasConcept C184720557 @default.
- W3209383860 hasConcept C199360897 @default.
- W3209383860 hasConcept C202444582 @default.
- W3209383860 hasConcept C2524010 @default.
- W3209383860 hasConcept C2778572836 @default.
- W3209383860 hasConcept C2780791683 @default.
- W3209383860 hasConcept C33923547 @default.
- W3209383860 hasConcept C34413123 @default.
- W3209383860 hasConcept C41008148 @default.
- W3209383860 hasConcept C48103436 @default.
- W3209383860 hasConcept C50644808 @default.
- W3209383860 hasConcept C62520636 @default.
- W3209383860 hasConcept C72434380 @default.
- W3209383860 hasConcept C75553542 @default.
- W3209383860 hasConcept C78458016 @default.
- W3209383860 hasConcept C80444323 @default.
- W3209383860 hasConcept C86803240 @default.
- W3209383860 hasConcept C90509273 @default.
- W3209383860 hasConcept C96469262 @default.
- W3209383860 hasConceptScore W3209383860C105795698 @default.
- W3209383860 hasConceptScore W3209383860C111472728 @default.
- W3209383860 hasConceptScore W3209383860C111919701 @default.
- W3209383860 hasConceptScore W3209383860C11413529 @default.
- W3209383860 hasConceptScore W3209383860C114614502 @default.
- W3209383860 hasConceptScore W3209383860C121332964 @default.
- W3209383860 hasConceptScore W3209383860C136197465 @default.
- W3209383860 hasConceptScore W3209383860C138885662 @default.
- W3209383860 hasConceptScore W3209383860C14036430 @default.
- W3209383860 hasConceptScore W3209383860C154945302 @default.
- W3209383860 hasConceptScore W3209383860C171036898 @default.