Matches in SemOpenAlex for { <https://semopenalex.org/work/W2892457101> ?p ?o ?g. }
- W2892457101 abstract "In many applications we seek to maximize an expectation with respect to a distribution over discrete variables. Estimating gradients of such objectives with respect to the distribution parameters is a challenging problem. We analyze existing solutions including finite-difference (FD) estimators and continuous relaxation (CR) estimators in terms of bias and variance. We show that the commonly used Gumbel-Softmax estimator is biased and propose a simple method to reduce it. We also derive a simpler piece-wise linear continuous relaxation that also possesses reduced bias. We demonstrate empirically that reduced bias leads to a better performance in variational inference and on binary optimization tasks." @default.
- W2892457101 created "2018-10-05" @default.
- W2892457101 creator A5038984764 @default.
- W2892457101 creator A5045431565 @default.
- W2892457101 creator A5076214952 @default.
- W2892457101 date "2018-09-29" @default.
- W2892457101 modified "2023-09-27" @default.
- W2892457101 title "Improved Gradient-Based Optimization Over Discrete Distributions." @default.
- W2892457101 cites W1522301498 @default.
- W2892457101 cites W1646152356 @default.
- W2892457101 cites W2046765929 @default.
- W2892457101 cites W2119717200 @default.
- W2892457101 cites W2187669537 @default.
- W2892457101 cites W2242818861 @default.
- W2892457101 cites W2259167532 @default.
- W2892457101 cites W2519430864 @default.
- W2892457101 cites W2602076750 @default.
- W2892457101 cites W2740840795 @default.
- W2892457101 cites W2767002724 @default.
- W2892457101 cites W2795629755 @default.
- W2892457101 cites W2806929479 @default.
- W2892457101 cites W2884043992 @default.
- W2892457101 cites W2909637611 @default.
- W2892457101 cites W2949595773 @default.
- W2892457101 cites W2950151997 @default.
- W2892457101 cites W2951004968 @default.
- W2892457101 cites W2952165242 @default.
- W2892457101 cites W2952264928 @default.
- W2892457101 cites W2962897886 @default.
- W2892457101 cites W2963223306 @default.
- W2892457101 cites W2963351648 @default.
- W2892457101 cites W2963408680 @default.
- W2892457101 cites W2963619462 @default.
- W2892457101 cites W2963812505 @default.
- W2892457101 hasPublicationYear "2018" @default.
- W2892457101 type Work @default.
- W2892457101 sameAs 2892457101 @default.
- W2892457101 citedByCount "3" @default.
- W2892457101 countsByYear W28924571012019 @default.
- W2892457101 countsByYear W28924571012020 @default.
- W2892457101 crossrefType "posted-content" @default.
- W2892457101 hasAuthorship W2892457101A5038984764 @default.
- W2892457101 hasAuthorship W2892457101A5045431565 @default.
- W2892457101 hasAuthorship W2892457101A5076214952 @default.
- W2892457101 hasConcept C105795698 @default.
- W2892457101 hasConcept C110121322 @default.
- W2892457101 hasConcept C111472728 @default.
- W2892457101 hasConcept C11413529 @default.
- W2892457101 hasConcept C121955636 @default.
- W2892457101 hasConcept C126255220 @default.
- W2892457101 hasConcept C134306372 @default.
- W2892457101 hasConcept C137610916 @default.
- W2892457101 hasConcept C138885662 @default.
- W2892457101 hasConcept C144133560 @default.
- W2892457101 hasConcept C147581598 @default.
- W2892457101 hasConcept C154945302 @default.
- W2892457101 hasConcept C15744967 @default.
- W2892457101 hasConcept C185429906 @default.
- W2892457101 hasConcept C196083921 @default.
- W2892457101 hasConcept C2776029896 @default.
- W2892457101 hasConcept C2776214188 @default.
- W2892457101 hasConcept C2780586882 @default.
- W2892457101 hasConcept C28826006 @default.
- W2892457101 hasConcept C33923547 @default.
- W2892457101 hasConcept C41008148 @default.
- W2892457101 hasConcept C48372109 @default.
- W2892457101 hasConcept C77805123 @default.
- W2892457101 hasConcept C94375191 @default.
- W2892457101 hasConceptScore W2892457101C105795698 @default.
- W2892457101 hasConceptScore W2892457101C110121322 @default.
- W2892457101 hasConceptScore W2892457101C111472728 @default.
- W2892457101 hasConceptScore W2892457101C11413529 @default.
- W2892457101 hasConceptScore W2892457101C121955636 @default.
- W2892457101 hasConceptScore W2892457101C126255220 @default.
- W2892457101 hasConceptScore W2892457101C134306372 @default.
- W2892457101 hasConceptScore W2892457101C137610916 @default.
- W2892457101 hasConceptScore W2892457101C138885662 @default.
- W2892457101 hasConceptScore W2892457101C144133560 @default.
- W2892457101 hasConceptScore W2892457101C147581598 @default.
- W2892457101 hasConceptScore W2892457101C154945302 @default.
- W2892457101 hasConceptScore W2892457101C15744967 @default.
- W2892457101 hasConceptScore W2892457101C185429906 @default.
- W2892457101 hasConceptScore W2892457101C196083921 @default.
- W2892457101 hasConceptScore W2892457101C2776029896 @default.
- W2892457101 hasConceptScore W2892457101C2776214188 @default.
- W2892457101 hasConceptScore W2892457101C2780586882 @default.
- W2892457101 hasConceptScore W2892457101C28826006 @default.
- W2892457101 hasConceptScore W2892457101C33923547 @default.
- W2892457101 hasConceptScore W2892457101C41008148 @default.
- W2892457101 hasConceptScore W2892457101C48372109 @default.
- W2892457101 hasConceptScore W2892457101C77805123 @default.
- W2892457101 hasConceptScore W2892457101C94375191 @default.
- W2892457101 hasLocation W28924571011 @default.
- W2892457101 hasOpenAccess W2892457101 @default.
- W2892457101 hasPrimaryLocation W28924571011 @default.
- W2892457101 hasRelatedWork W1515046342 @default.
- W2892457101 hasRelatedWork W154712720 @default.
- W2892457101 hasRelatedWork W1768201245 @default.
- W2892457101 hasRelatedWork W1865551748 @default.
- W2892457101 hasRelatedWork W1972274727 @default.