Matches in SemOpenAlex for { <https://semopenalex.org/work/W4310266070> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W4310266070 abstract "Optimizing functions without access to gradients is the remit of black-box methods such as evolution strategies. While highly general, their learning dynamics are often times heuristic and inflexible - exactly the limitations that meta-learning can address. Hence, we propose to discover effective update rules for evolution strategies via meta-learning. Concretely, our approach employs a search strategy parametrized by a self-attention-based architecture, which guarantees the update rule is invariant to the ordering of the candidate solutions. We show that meta-evolving this system on a small set of representative low-dimensional analytic optimization problems is sufficient to discover new evolution strategies capable of generalizing to unseen optimization problems, population sizes and optimization horizons. Furthermore, the same learned evolution strategy can outperform established neuroevolution baselines on supervised and continuous control tasks. As additional contributions, we ablate the individual neural network components of our method; reverse engineer the learned strategy into an explicit heuristic form, which remains highly competitive; and show that it is possible to self-referentially train an evolution strategy from scratch, with the learned update rule used to drive the outer meta-learning loop." @default.
- W4310266070 created "2022-11-30" @default.
- W4310266070 creator A5001118898 @default.
- W4310266070 creator A5013857977 @default.
- W4310266070 creator A5018613019 @default.
- W4310266070 creator A5030505775 @default.
- W4310266070 creator A5043624160 @default.
- W4310266070 creator A5065366930 @default.
- W4310266070 creator A5074021977 @default.
- W4310266070 creator A5081322018 @default.
- W4310266070 date "2022-11-21" @default.
- W4310266070 modified "2023-09-25" @default.
- W4310266070 title "Discovering Evolution Strategies via Meta-Black-Box Optimization" @default.
- W4310266070 doi "https://doi.org/10.48550/arxiv.2211.11260" @default.
- W4310266070 hasPublicationYear "2022" @default.
- W4310266070 type Work @default.
- W4310266070 citedByCount "0" @default.
- W4310266070 crossrefType "posted-content" @default.
- W4310266070 hasAuthorship W4310266070A5001118898 @default.
- W4310266070 hasAuthorship W4310266070A5013857977 @default.
- W4310266070 hasAuthorship W4310266070A5018613019 @default.
- W4310266070 hasAuthorship W4310266070A5030505775 @default.
- W4310266070 hasAuthorship W4310266070A5043624160 @default.
- W4310266070 hasAuthorship W4310266070A5065366930 @default.
- W4310266070 hasAuthorship W4310266070A5074021977 @default.
- W4310266070 hasAuthorship W4310266070A5081322018 @default.
- W4310266070 hasBestOaLocation W43102660701 @default.
- W4310266070 hasConcept C11413529 @default.
- W4310266070 hasConcept C118070581 @default.
- W4310266070 hasConcept C119857082 @default.
- W4310266070 hasConcept C126255220 @default.
- W4310266070 hasConcept C127413603 @default.
- W4310266070 hasConcept C137836250 @default.
- W4310266070 hasConcept C144024400 @default.
- W4310266070 hasConcept C149923435 @default.
- W4310266070 hasConcept C154945302 @default.
- W4310266070 hasConcept C159149176 @default.
- W4310266070 hasConcept C173801870 @default.
- W4310266070 hasConcept C177264268 @default.
- W4310266070 hasConcept C199360897 @default.
- W4310266070 hasConcept C201995342 @default.
- W4310266070 hasConcept C207002847 @default.
- W4310266070 hasConcept C2780451532 @default.
- W4310266070 hasConcept C2781002164 @default.
- W4310266070 hasConcept C2908647359 @default.
- W4310266070 hasConcept C33923547 @default.
- W4310266070 hasConcept C41008148 @default.
- W4310266070 hasConcept C50644808 @default.
- W4310266070 hasConcept C94966114 @default.
- W4310266070 hasConceptScore W4310266070C11413529 @default.
- W4310266070 hasConceptScore W4310266070C118070581 @default.
- W4310266070 hasConceptScore W4310266070C119857082 @default.
- W4310266070 hasConceptScore W4310266070C126255220 @default.
- W4310266070 hasConceptScore W4310266070C127413603 @default.
- W4310266070 hasConceptScore W4310266070C137836250 @default.
- W4310266070 hasConceptScore W4310266070C144024400 @default.
- W4310266070 hasConceptScore W4310266070C149923435 @default.
- W4310266070 hasConceptScore W4310266070C154945302 @default.
- W4310266070 hasConceptScore W4310266070C159149176 @default.
- W4310266070 hasConceptScore W4310266070C173801870 @default.
- W4310266070 hasConceptScore W4310266070C177264268 @default.
- W4310266070 hasConceptScore W4310266070C199360897 @default.
- W4310266070 hasConceptScore W4310266070C201995342 @default.
- W4310266070 hasConceptScore W4310266070C207002847 @default.
- W4310266070 hasConceptScore W4310266070C2780451532 @default.
- W4310266070 hasConceptScore W4310266070C2781002164 @default.
- W4310266070 hasConceptScore W4310266070C2908647359 @default.
- W4310266070 hasConceptScore W4310266070C33923547 @default.
- W4310266070 hasConceptScore W4310266070C41008148 @default.
- W4310266070 hasConceptScore W4310266070C50644808 @default.
- W4310266070 hasConceptScore W4310266070C94966114 @default.
- W4310266070 hasLocation W43102660701 @default.
- W4310266070 hasOpenAccess W4310266070 @default.
- W4310266070 hasPrimaryLocation W43102660701 @default.
- W4310266070 hasRelatedWork W3020638616 @default.
- W4310266070 hasRelatedWork W3092824172 @default.
- W4310266070 hasRelatedWork W3105036711 @default.
- W4310266070 hasRelatedWork W3199608561 @default.
- W4310266070 hasRelatedWork W3200361725 @default.
- W4310266070 hasRelatedWork W4213165337 @default.
- W4310266070 hasRelatedWork W4225154779 @default.
- W4310266070 hasRelatedWork W4307291644 @default.
- W4310266070 hasRelatedWork W4310266070 @default.
- W4310266070 hasRelatedWork W1629725936 @default.
- W4310266070 isParatext "false" @default.
- W4310266070 isRetracted "false" @default.
- W4310266070 workType "article" @default.