Matches in SemOpenAlex for { <https://semopenalex.org/work/W4304195540> ?p ?o ?g. }
Showing items 1 to 67 of
67
with 100 items per page.
- W4304195540 abstract "Reinforcement learning (RL) agents have long sought to approach the efficiency of human learning. Humans are great observers who can learn by aggregating external knowledge from various sources, including observations from others' policies of attempting a task. Prior studies in RL have incorporated external knowledge policies to help agents improve sample efficiency. However, it remains non-trivial to perform arbitrary combinations and replacements of those policies, an essential feature for generalization and transferability. In this work, we present Knowledge-Grounded RL (KGRL), an RL paradigm fusing multiple knowledge policies and aiming for human-like efficiency and flexibility. We propose a new actor architecture for KGRL, Knowledge-Inclusive Attention Network (KIAN), which allows free knowledge rearrangement due to embedding-based attentive action prediction. KIAN also addresses entropy imbalance, a problem arising in maximum entropy KGRL that hinders an agent from efficiently exploring the environment, through a new design of policy distributions. The experimental results demonstrate that KIAN outperforms alternative methods incorporating external knowledge policies and achieves efficient and flexible learning. Our implementation is available at https://github.com/Pascalson/KGRL.git" @default.
- W4304195540 created "2022-10-11" @default.
- W4304195540 creator A5016343106 @default.
- W4304195540 creator A5050195037 @default.
- W4304195540 creator A5054598974 @default.
- W4304195540 creator A5077633096 @default.
- W4304195540 date "2022-10-07" @default.
- W4304195540 modified "2023-10-13" @default.
- W4304195540 title "Flexible Attention-Based Multi-Policy Fusion for Efficient Deep Reinforcement Learning" @default.
- W4304195540 doi "https://doi.org/10.48550/arxiv.2210.03729" @default.
- W4304195540 hasPublicationYear "2022" @default.
- W4304195540 type Work @default.
- W4304195540 citedByCount "0" @default.
- W4304195540 crossrefType "posted-content" @default.
- W4304195540 hasAuthorship W4304195540A5016343106 @default.
- W4304195540 hasAuthorship W4304195540A5050195037 @default.
- W4304195540 hasAuthorship W4304195540A5054598974 @default.
- W4304195540 hasAuthorship W4304195540A5077633096 @default.
- W4304195540 hasBestOaLocation W43041955401 @default.
- W4304195540 hasConcept C105795698 @default.
- W4304195540 hasConcept C106301342 @default.
- W4304195540 hasConcept C119857082 @default.
- W4304195540 hasConcept C121332964 @default.
- W4304195540 hasConcept C134306372 @default.
- W4304195540 hasConcept C140331021 @default.
- W4304195540 hasConcept C154945302 @default.
- W4304195540 hasConcept C177148314 @default.
- W4304195540 hasConcept C2779436431 @default.
- W4304195540 hasConcept C2780598303 @default.
- W4304195540 hasConcept C33923547 @default.
- W4304195540 hasConcept C41008148 @default.
- W4304195540 hasConcept C41608201 @default.
- W4304195540 hasConcept C61272859 @default.
- W4304195540 hasConcept C62520636 @default.
- W4304195540 hasConcept C97541855 @default.
- W4304195540 hasConceptScore W4304195540C105795698 @default.
- W4304195540 hasConceptScore W4304195540C106301342 @default.
- W4304195540 hasConceptScore W4304195540C119857082 @default.
- W4304195540 hasConceptScore W4304195540C121332964 @default.
- W4304195540 hasConceptScore W4304195540C134306372 @default.
- W4304195540 hasConceptScore W4304195540C140331021 @default.
- W4304195540 hasConceptScore W4304195540C154945302 @default.
- W4304195540 hasConceptScore W4304195540C177148314 @default.
- W4304195540 hasConceptScore W4304195540C2779436431 @default.
- W4304195540 hasConceptScore W4304195540C2780598303 @default.
- W4304195540 hasConceptScore W4304195540C33923547 @default.
- W4304195540 hasConceptScore W4304195540C41008148 @default.
- W4304195540 hasConceptScore W4304195540C41608201 @default.
- W4304195540 hasConceptScore W4304195540C61272859 @default.
- W4304195540 hasConceptScore W4304195540C62520636 @default.
- W4304195540 hasConceptScore W4304195540C97541855 @default.
- W4304195540 hasLocation W43041955401 @default.
- W4304195540 hasOpenAccess W4304195540 @default.
- W4304195540 hasPrimaryLocation W43041955401 @default.
- W4304195540 hasRelatedWork W2011110943 @default.
- W4304195540 hasRelatedWork W2011433332 @default.
- W4304195540 hasRelatedWork W2055748329 @default.
- W4304195540 hasRelatedWork W2161221533 @default.
- W4304195540 hasRelatedWork W2216382288 @default.
- W4304195540 hasRelatedWork W2355491300 @default.
- W4304195540 hasRelatedWork W2582594227 @default.
- W4304195540 hasRelatedWork W4234629551 @default.
- W4304195540 hasRelatedWork W1534779234 @default.
- W4304195540 hasRelatedWork W1666484574 @default.
- W4304195540 isParatext "false" @default.
- W4304195540 isRetracted "false" @default.
- W4304195540 workType "article" @default.