Matches in SemOpenAlex for { <https://semopenalex.org/work/W2789776893> ?p ?o ?g. }
- W2789776893 abstract "Recent work in unsupervised representation learning has focused on learning deep directed latent-variable models. Fitting these models by maximizing the marginal likelihood or evidence is typically intractable, thus a common approximation is to maximize the evidence lower bound (ELBO) instead. However, maximum likelihood training (whether exact or approximate) does not necessarily result in a good latent representation, as we demonstrate both theoretically and empirically. In particular, we derive variational lower and upper bounds on the mutual information between the input and the latent variable, and use these bounds to derive a rate-distortion curve that characterizes the tradeoff between compression and reconstruction accuracy. Using this framework, we demonstrate that there is a family of models with identical ELBO, but different quantitative and qualitative characteristics. Our framework also suggests a simple new method to ensure that latent variable models with powerful stochastic decoders do not ignore their latent code." @default.
- W2789776893 created "2018-03-29" @default.
- W2789776893 creator A5002713363 @default.
- W2789776893 creator A5004918186 @default.
- W2789776893 creator A5021577029 @default.
- W2789776893 creator A5026650852 @default.
- W2789776893 creator A5036325834 @default.
- W2789776893 creator A5051461147 @default.
- W2789776893 date "2017-11-01" @default.
- W2789776893 modified "2023-09-27" @default.
- W2789776893 title "Fixing a Broken ELBO" @default.
- W2789776893 cites W1580389772 @default.
- W2789776893 cites W1686946872 @default.
- W2789776893 cites W1779483307 @default.
- W2789776893 cites W1909320841 @default.
- W2789776893 cites W1936878994 @default.
- W2789776893 cites W1959608418 @default.
- W2789776893 cites W2047229728 @default.
- W2789776893 cites W2114771311 @default.
- W2789776893 cites W2135181320 @default.
- W2789776893 cites W2194321275 @default.
- W2789776893 cites W2210838531 @default.
- W2789776893 cites W2228605760 @default.
- W2789776893 cites W2274287116 @default.
- W2789776893 cites W2395149821 @default.
- W2789776893 cites W2397262406 @default.
- W2789776893 cites W2399306074 @default.
- W2789776893 cites W2423557781 @default.
- W2789776893 cites W2431962807 @default.
- W2789776893 cites W2439880944 @default.
- W2789776893 cites W2552465432 @default.
- W2789776893 cites W2557579533 @default.
- W2789776893 cites W2557728737 @default.
- W2789776893 cites W2593634001 @default.
- W2789776893 cites W2616159609 @default.
- W2789776893 cites W2617620476 @default.
- W2789776893 cites W2619016545 @default.
- W2789776893 cites W2626778328 @default.
- W2789776893 cites W2683470288 @default.
- W2789776893 cites W2737818580 @default.
- W2789776893 cites W2753738274 @default.
- W2789776893 cites W2809994596 @default.
- W2789776893 cites W2949899814 @default.
- W2789776893 cites W2950220847 @default.
- W2789776893 cites W2950237263 @default.
- W2789776893 cites W2952830092 @default.
- W2789776893 cites W2952838738 @default.
- W2789776893 cites W2953046278 @default.
- W2789776893 cites W2963970792 @default.
- W2789776893 cites W2964121744 @default.
- W2789776893 cites W2964122153 @default.
- W2789776893 cites W299440670 @default.
- W2789776893 hasPublicationYear "2017" @default.
- W2789776893 type Work @default.
- W2789776893 sameAs 2789776893 @default.
- W2789776893 citedByCount "41" @default.
- W2789776893 countsByYear W27897768932017 @default.
- W2789776893 countsByYear W27897768932018 @default.
- W2789776893 countsByYear W27897768932019 @default.
- W2789776893 countsByYear W27897768932020 @default.
- W2789776893 countsByYear W27897768932021 @default.
- W2789776893 crossrefType "posted-content" @default.
- W2789776893 hasAuthorship W2789776893A5002713363 @default.
- W2789776893 hasAuthorship W2789776893A5004918186 @default.
- W2789776893 hasAuthorship W2789776893A5021577029 @default.
- W2789776893 hasAuthorship W2789776893A5026650852 @default.
- W2789776893 hasAuthorship W2789776893A5036325834 @default.
- W2789776893 hasAuthorship W2789776893A5051461147 @default.
- W2789776893 hasConcept C107673813 @default.
- W2789776893 hasConcept C111472728 @default.
- W2789776893 hasConcept C11413529 @default.
- W2789776893 hasConcept C119857082 @default.
- W2789776893 hasConcept C126780896 @default.
- W2789776893 hasConcept C134306372 @default.
- W2789776893 hasConcept C138885662 @default.
- W2789776893 hasConcept C154945302 @default.
- W2789776893 hasConcept C177264268 @default.
- W2789776893 hasConcept C17744445 @default.
- W2789776893 hasConcept C182365436 @default.
- W2789776893 hasConcept C194257627 @default.
- W2789776893 hasConcept C199360897 @default.
- W2789776893 hasConcept C199539241 @default.
- W2789776893 hasConcept C2776257435 @default.
- W2789776893 hasConcept C2776359362 @default.
- W2789776893 hasConcept C2776760102 @default.
- W2789776893 hasConcept C2780586882 @default.
- W2789776893 hasConcept C31258907 @default.
- W2789776893 hasConcept C33923547 @default.
- W2789776893 hasConcept C41008148 @default.
- W2789776893 hasConcept C51167844 @default.
- W2789776893 hasConcept C65965080 @default.
- W2789776893 hasConcept C77553402 @default.
- W2789776893 hasConcept C94625758 @default.
- W2789776893 hasConcept C95923904 @default.
- W2789776893 hasConceptScore W2789776893C107673813 @default.
- W2789776893 hasConceptScore W2789776893C111472728 @default.
- W2789776893 hasConceptScore W2789776893C11413529 @default.
- W2789776893 hasConceptScore W2789776893C119857082 @default.
- W2789776893 hasConceptScore W2789776893C126780896 @default.
- W2789776893 hasConceptScore W2789776893C134306372 @default.