FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3042380218> ?p ?o ?g. }

• W3042380218 abstract "Batch reinforcement learning (RL) is important to apply RL algorithms to many high stakes tasks. Doing batch RL in a way that yields a reliable new policy in large domains is challenging: a new decision policy may visit states and actions outside the support of the batch data, and function approximation and optimization with limited samples can further increase the potential of learning policies with overly optimistic estimates of their future performance. Recent algorithms have shown promise but can still be overly optimistic in their expected outcomes. Theoretical work that provides strong guarantees on the performance of the output policy relies on a strong concentrability assumption, that makes it unsuitable for cases where the ratio between state-action distributions of behavior policy and some candidate policies is large. This is because in the traditional analysis, the error bound scales up with this ratio. We show that a small modification to Bellman optimality and evaluation back-up to take a more conservative update can have much stronger guarantees. In certain settings, they can find the approximately best policy within the state-action space explored by the batch data, without requiring a priori assumptions of concentrability. We highlight the necessity of our conservative update and the limitations of previous algorithms and analyses by illustrative MDP examples, and demonstrate an empirical comparison of our algorithm and other state-of-the-art batch RL baselines in standard benchmarks." @default.
• W3042380218 created "2020-07-23" @default.
• W3042380218 creator A5034711562 @default.
• W3042380218 creator A5036435487 @default.
• W3042380218 creator A5066715218 @default.
• W3042380218 creator A5084989076 @default.
• W3042380218 date "2020-07-16" @default.
• W3042380218 modified "2023-09-26" @default.
• W3042380218 title "Provably Good Batch Reinforcement Learning Without Great Exploration." @default.
• W3042380218 cites W114364418 @default.
• W3042380218 cites W1492095942 @default.
• W3042380218 cites W1575592356 @default.
• W3042380218 cites W166862392 @default.
• W3042380218 cites W1730555343 @default.
• W3042380218 cites W1825869920 @default.
• W3042380218 cites W192920577 @default.
• W3042380218 cites W1931027396 @default.
• W3042380218 cites W1995514448 @default.
• W3042380218 cites W2041277185 @default.
• W3042380218 cites W2086333522 @default.
• W3042380218 cites W2104753538 @default.
• W3042380218 cites W2117355432 @default.
• W3042380218 cites W2119579400 @default.
• W3042380218 cites W2120346334 @default.
• W3042380218 cites W2121863487 @default.
• W3042380218 cites W2122689259 @default.
• W3042380218 cites W2124477018 @default.
• W3042380218 cites W2128812357 @default.
• W3042380218 cites W2130005627 @default.
• W3042380218 cites W2130599357 @default.
• W3042380218 cites W2130801532 @default.
• W3042380218 cites W2141559645 @default.
• W3042380218 cites W2145339207 @default.
• W3042380218 cites W2460675832 @default.
• W3042380218 cites W2520501711 @default.
• W3042380218 cites W2949257336 @default.
• W3042380218 cites W2952500758 @default.
• W3042380218 cites W2962734844 @default.
• W3042380218 cites W2962785728 @default.
• W3042380218 cites W2963704132 @default.
• W3042380218 cites W2971262355 @default.
• W3042380218 cites W2991598122 @default.
• W3042380218 cites W3009962997 @default.
• W3042380218 cites W3034607397 @default.
• W3042380218 cites W3039845099 @default.
• W3042380218 cites W3046626913 @default.
• W3042380218 cites W3093206925 @default.
• W3042380218 hasPublicationYear "2020" @default.
• W3042380218 type Work @default.
• W3042380218 sameAs 3042380218 @default.
• W3042380218 citedByCount "29" @default.
• W3042380218 countsByYear W30423802182020 @default.
• W3042380218 countsByYear W30423802182021 @default.
• W3042380218 crossrefType "posted-content" @default.
• W3042380218 hasAuthorship W3042380218A5034711562 @default.
• W3042380218 hasAuthorship W3042380218A5036435487 @default.
• W3042380218 hasAuthorship W3042380218A5066715218 @default.
• W3042380218 hasAuthorship W3042380218A5084989076 @default.
• W3042380218 hasConcept C105795698 @default.
• W3042380218 hasConcept C111472728 @default.
• W3042380218 hasConcept C111919701 @default.
• W3042380218 hasConcept C11413529 @default.
• W3042380218 hasConcept C119857082 @default.
• W3042380218 hasConcept C121332964 @default.
• W3042380218 hasConcept C126255220 @default.
• W3042380218 hasConcept C138885662 @default.
• W3042380218 hasConcept C14036430 @default.
• W3042380218 hasConcept C154945302 @default.
• W3042380218 hasConcept C2778572836 @default.
• W3042380218 hasConcept C2780791683 @default.
• W3042380218 hasConcept C33923547 @default.
• W3042380218 hasConcept C41008148 @default.
• W3042380218 hasConcept C48103436 @default.
• W3042380218 hasConcept C62520636 @default.
• W3042380218 hasConcept C72434380 @default.
• W3042380218 hasConcept C75553542 @default.
• W3042380218 hasConcept C78458016 @default.
• W3042380218 hasConcept C86803240 @default.
• W3042380218 hasConcept C97541855 @default.
• W3042380218 hasConceptScore W3042380218C105795698 @default.
• W3042380218 hasConceptScore W3042380218C111472728 @default.
• W3042380218 hasConceptScore W3042380218C111919701 @default.
• W3042380218 hasConceptScore W3042380218C11413529 @default.
• W3042380218 hasConceptScore W3042380218C119857082 @default.
• W3042380218 hasConceptScore W3042380218C121332964 @default.
• W3042380218 hasConceptScore W3042380218C126255220 @default.
• W3042380218 hasConceptScore W3042380218C138885662 @default.
• W3042380218 hasConceptScore W3042380218C14036430 @default.
• W3042380218 hasConceptScore W3042380218C154945302 @default.
• W3042380218 hasConceptScore W3042380218C2778572836 @default.
• W3042380218 hasConceptScore W3042380218C2780791683 @default.
• W3042380218 hasConceptScore W3042380218C33923547 @default.
• W3042380218 hasConceptScore W3042380218C41008148 @default.
• W3042380218 hasConceptScore W3042380218C48103436 @default.
• W3042380218 hasConceptScore W3042380218C62520636 @default.
• W3042380218 hasConceptScore W3042380218C72434380 @default.
• W3042380218 hasConceptScore W3042380218C75553542 @default.
• W3042380218 hasConceptScore W3042380218C78458016 @default.
• W3042380218 hasConceptScore W3042380218C86803240 @default.
• W3042380218 hasConceptScore W3042380218C97541855 @default.

• next