Matches in SemOpenAlex for { <https://semopenalex.org/work/W3123322590> ?p ?o ?g. }
Showing items 1 to 93 of
93
with 100 items per page.
- W3123322590 abstract "This work explores learning agent-agnostic synthetic environments (SEs) for Reinforcement Learning. SEs act as a proxy for target environments and allow agents to be trained more efficiently than when directly trained on the target environment. We formulate this as a bi-level optimization problem and represent an SE as a neural network. By using Natural Evolution Strategies and a population of SE parameter vectors, we train agents in the inner loop on evolving SEs while in the outer loop we use the performance on the target task as a score for meta-updating the SE population. We show empirically that our method is capable of learning SEs for two discrete-action-space tasks (CartPole-v0 and Acrobot-v1) that allow us to train agents more robustly and with up to 60% fewer steps. Not only do we show in experiments with 4000 evaluations that the SEs are robust against hyperparameter changes such as the learning rate, batch sizes and network sizes, we also show that SEs trained with DDQN agents transfer in limited ways to a discrete-action-space version of TD3 and very well to Dueling DDQN." @default.
- W3123322590 created "2021-02-01" @default.
- W3123322590 creator A5031002895 @default.
- W3123322590 creator A5076691711 @default.
- W3123322590 creator A5088293505 @default.
- W3123322590 date "2021-01-24" @default.
- W3123322590 modified "2023-09-27" @default.
- W3123322590 title "Learning Synthetic Environments for Reinforcement Learning with Evolution Strategies." @default.
- W3123322590 cites W1506498113 @default.
- W3123322590 cites W1514875444 @default.
- W3123322590 cites W2155027007 @default.
- W3123322590 cites W2155968351 @default.
- W3123322590 cites W2166160300 @default.
- W3123322590 cites W2173564293 @default.
- W3123322590 cites W2547875792 @default.
- W3123322590 cites W2596367596 @default.
- W3123322590 cites W2795843265 @default.
- W3123322590 cites W2908460759 @default.
- W3123322590 cites W2916158460 @default.
- W3123322590 cites W2963423218 @default.
- W3123322590 cites W2963902936 @default.
- W3123322590 cites W2963923407 @default.
- W3123322590 cites W2964096423 @default.
- W3123322590 cites W3005596810 @default.
- W3123322590 cites W3011120880 @default.
- W3123322590 cites W3034828529 @default.
- W3123322590 cites W3037207827 @default.
- W3123322590 hasPublicationYear "2021" @default.
- W3123322590 type Work @default.
- W3123322590 sameAs 3123322590 @default.
- W3123322590 citedByCount "1" @default.
- W3123322590 countsByYear W31233225902021 @default.
- W3123322590 crossrefType "posted-content" @default.
- W3123322590 hasAuthorship W3123322590A5031002895 @default.
- W3123322590 hasAuthorship W3123322590A5076691711 @default.
- W3123322590 hasAuthorship W3123322590A5088293505 @default.
- W3123322590 hasConcept C111919701 @default.
- W3123322590 hasConcept C119857082 @default.
- W3123322590 hasConcept C127413603 @default.
- W3123322590 hasConcept C144024400 @default.
- W3123322590 hasConcept C149923435 @default.
- W3123322590 hasConcept C150899416 @default.
- W3123322590 hasConcept C154945302 @default.
- W3123322590 hasConcept C201995342 @default.
- W3123322590 hasConcept C2778572836 @default.
- W3123322590 hasConcept C2780451532 @default.
- W3123322590 hasConcept C2908647359 @default.
- W3123322590 hasConcept C41008148 @default.
- W3123322590 hasConcept C50644808 @default.
- W3123322590 hasConcept C8642999 @default.
- W3123322590 hasConcept C97541855 @default.
- W3123322590 hasConceptScore W3123322590C111919701 @default.
- W3123322590 hasConceptScore W3123322590C119857082 @default.
- W3123322590 hasConceptScore W3123322590C127413603 @default.
- W3123322590 hasConceptScore W3123322590C144024400 @default.
- W3123322590 hasConceptScore W3123322590C149923435 @default.
- W3123322590 hasConceptScore W3123322590C150899416 @default.
- W3123322590 hasConceptScore W3123322590C154945302 @default.
- W3123322590 hasConceptScore W3123322590C201995342 @default.
- W3123322590 hasConceptScore W3123322590C2778572836 @default.
- W3123322590 hasConceptScore W3123322590C2780451532 @default.
- W3123322590 hasConceptScore W3123322590C2908647359 @default.
- W3123322590 hasConceptScore W3123322590C41008148 @default.
- W3123322590 hasConceptScore W3123322590C50644808 @default.
- W3123322590 hasConceptScore W3123322590C8642999 @default.
- W3123322590 hasConceptScore W3123322590C97541855 @default.
- W3123322590 hasLocation W31233225901 @default.
- W3123322590 hasOpenAccess W3123322590 @default.
- W3123322590 hasPrimaryLocation W31233225901 @default.
- W3123322590 hasRelatedWork W2112422203 @default.
- W3123322590 hasRelatedWork W2160308170 @default.
- W3123322590 hasRelatedWork W2166265228 @default.
- W3123322590 hasRelatedWork W2289410116 @default.
- W3123322590 hasRelatedWork W2494907211 @default.
- W3123322590 hasRelatedWork W2528846071 @default.
- W3123322590 hasRelatedWork W2765297602 @default.
- W3123322590 hasRelatedWork W2804672169 @default.
- W3123322590 hasRelatedWork W2962817122 @default.
- W3123322590 hasRelatedWork W2967645217 @default.
- W3123322590 hasRelatedWork W2989988749 @default.
- W3123322590 hasRelatedWork W2990911039 @default.
- W3123322590 hasRelatedWork W3029221344 @default.
- W3123322590 hasRelatedWork W3037940279 @default.
- W3123322590 hasRelatedWork W3042593733 @default.
- W3123322590 hasRelatedWork W3131944163 @default.
- W3123322590 hasRelatedWork W3196548425 @default.
- W3123322590 hasRelatedWork W3198965389 @default.
- W3123322590 hasRelatedWork W3205046940 @default.
- W3123322590 hasRelatedWork W3209208698 @default.
- W3123322590 isParatext "false" @default.
- W3123322590 isRetracted "false" @default.
- W3123322590 magId "3123322590" @default.
- W3123322590 workType "article" @default.