Matches in SemOpenAlex for { <https://semopenalex.org/work/W4383047441> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W4383047441 abstract "We study the convergence behavior of the celebrated temporal-difference (TD) learning algorithm. By looking at the algorithm through the lens of optimization, we first argue that TD can be viewed as an iterative optimization algorithm where the function to be minimized changes per iteration. By carefully investigating the divergence displayed by TD on a classical counter example, we identify two forces that determine the convergent or divergent behavior of the algorithm. We next formalize our discovery in the linear TD setting with quadratic loss and prove that convergence of TD hinges on the interplay between these two forces. We extend this optimization perspective to prove convergence of TD in a much broader setting than just linear approximation and squared loss. Our results provide a theoretical explanation for the successful application of TD in reinforcement learning." @default.
- W4383047441 created "2023-07-04" @default.
- W4383047441 creator A5007854617 @default.
- W4383047441 creator A5010753342 @default.
- W4383047441 creator A5037667167 @default.
- W4383047441 creator A5045518550 @default.
- W4383047441 creator A5091908162 @default.
- W4383047441 date "2023-06-30" @default.
- W4383047441 modified "2023-09-26" @default.
- W4383047441 title "TD Convergence: An Optimization Perspective" @default.
- W4383047441 doi "https://doi.org/10.48550/arxiv.2306.17750" @default.
- W4383047441 hasPublicationYear "2023" @default.
- W4383047441 type Work @default.
- W4383047441 citedByCount "0" @default.
- W4383047441 crossrefType "posted-content" @default.
- W4383047441 hasAuthorship W4383047441A5007854617 @default.
- W4383047441 hasAuthorship W4383047441A5010753342 @default.
- W4383047441 hasAuthorship W4383047441A5037667167 @default.
- W4383047441 hasAuthorship W4383047441A5045518550 @default.
- W4383047441 hasAuthorship W4383047441A5091908162 @default.
- W4383047441 hasBestOaLocation W43830474411 @default.
- W4383047441 hasConcept C11413529 @default.
- W4383047441 hasConcept C12267149 @default.
- W4383047441 hasConcept C126255220 @default.
- W4383047441 hasConcept C12713177 @default.
- W4383047441 hasConcept C129844170 @default.
- W4383047441 hasConcept C138885662 @default.
- W4383047441 hasConcept C14036430 @default.
- W4383047441 hasConcept C154945302 @default.
- W4383047441 hasConcept C162324750 @default.
- W4383047441 hasConcept C178635117 @default.
- W4383047441 hasConcept C207390915 @default.
- W4383047441 hasConcept C2524010 @default.
- W4383047441 hasConcept C2777303404 @default.
- W4383047441 hasConcept C28826006 @default.
- W4383047441 hasConcept C33923547 @default.
- W4383047441 hasConcept C38652104 @default.
- W4383047441 hasConcept C39891107 @default.
- W4383047441 hasConcept C41008148 @default.
- W4383047441 hasConcept C41045048 @default.
- W4383047441 hasConcept C41895202 @default.
- W4383047441 hasConcept C50522688 @default.
- W4383047441 hasConcept C78458016 @default.
- W4383047441 hasConcept C86803240 @default.
- W4383047441 hasConcept C89109886 @default.
- W4383047441 hasConcept C97541855 @default.
- W4383047441 hasConceptScore W4383047441C11413529 @default.
- W4383047441 hasConceptScore W4383047441C12267149 @default.
- W4383047441 hasConceptScore W4383047441C126255220 @default.
- W4383047441 hasConceptScore W4383047441C12713177 @default.
- W4383047441 hasConceptScore W4383047441C129844170 @default.
- W4383047441 hasConceptScore W4383047441C138885662 @default.
- W4383047441 hasConceptScore W4383047441C14036430 @default.
- W4383047441 hasConceptScore W4383047441C154945302 @default.
- W4383047441 hasConceptScore W4383047441C162324750 @default.
- W4383047441 hasConceptScore W4383047441C178635117 @default.
- W4383047441 hasConceptScore W4383047441C207390915 @default.
- W4383047441 hasConceptScore W4383047441C2524010 @default.
- W4383047441 hasConceptScore W4383047441C2777303404 @default.
- W4383047441 hasConceptScore W4383047441C28826006 @default.
- W4383047441 hasConceptScore W4383047441C33923547 @default.
- W4383047441 hasConceptScore W4383047441C38652104 @default.
- W4383047441 hasConceptScore W4383047441C39891107 @default.
- W4383047441 hasConceptScore W4383047441C41008148 @default.
- W4383047441 hasConceptScore W4383047441C41045048 @default.
- W4383047441 hasConceptScore W4383047441C41895202 @default.
- W4383047441 hasConceptScore W4383047441C50522688 @default.
- W4383047441 hasConceptScore W4383047441C78458016 @default.
- W4383047441 hasConceptScore W4383047441C86803240 @default.
- W4383047441 hasConceptScore W4383047441C89109886 @default.
- W4383047441 hasConceptScore W4383047441C97541855 @default.
- W4383047441 hasLocation W43830474411 @default.
- W4383047441 hasOpenAccess W4383047441 @default.
- W4383047441 hasPrimaryLocation W43830474411 @default.
- W4383047441 hasRelatedWork W1820996291 @default.
- W4383047441 hasRelatedWork W2066217924 @default.
- W4383047441 hasRelatedWork W2071310731 @default.
- W4383047441 hasRelatedWork W2102267274 @default.
- W4383047441 hasRelatedWork W2348351659 @default.
- W4383047441 hasRelatedWork W2351037799 @default.
- W4383047441 hasRelatedWork W2379005117 @default.
- W4383047441 hasRelatedWork W4361253176 @default.
- W4383047441 hasRelatedWork W4380302820 @default.
- W4383047441 hasRelatedWork W2745007822 @default.
- W4383047441 isParatext "false" @default.
- W4383047441 isRetracted "false" @default.
- W4383047441 workType "article" @default.