Matches in SemOpenAlex for { <https://semopenalex.org/work/W2808508258> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W2808508258 abstract "Powerful generative models, particularly in Natural Language Modelling, are commonly trained by maximizing a variational lower bound on the data log likelihood. These models often suffer from poor use of their latent variable, with ad-hoc annealing factors used to encourage retention of information in the latent variable. We discuss an alternative and general approach to latent variable modelling, based on an objective that combines the data log likelihood as well as the likelihood of a perfect reconstruction through an autoencoder. Tying these together ensures by design that the latent variable captures information about the observations, whilst retaining the ability to generate well. Interestingly, though this approach is a priori unrelated to VAEs, the lower bound attained is identical to the standard VAE bound but with the addition of a simple pre-factor; thus, providing a formal interpretation of the commonly used, ad-hoc pre-factors in training VAEs." @default.
- W2808508258 created "2018-06-21" @default.
- W2808508258 creator A5022469843 @default.
- W2808508258 creator A5053299181 @default.
- W2808508258 creator A5063260720 @default.
- W2808508258 creator A5074723086 @default.
- W2808508258 date "2018-06-12" @default.
- W2808508258 modified "2023-09-27" @default.
- W2808508258 title "Improving latent variable descriptiveness with AutoGen." @default.
- W2808508258 cites W1909320841 @default.
- W2808508258 cites W1959608418 @default.
- W2808508258 cites W2171361956 @default.
- W2808508258 cites W2560512785 @default.
- W2808508258 cites W2594538354 @default.
- W2808508258 cites W2753738274 @default.
- W2808508258 cites W2788982938 @default.
- W2808508258 cites W2949433733 @default.
- W2808508258 cites W2949899814 @default.
- W2808508258 cites W2951326654 @default.
- W2808508258 cites W2951575317 @default.
- W2808508258 cites W2952723479 @default.
- W2808508258 cites W2963143316 @default.
- W2808508258 cites W2963223306 @default.
- W2808508258 hasPublicationYear "2018" @default.
- W2808508258 type Work @default.
- W2808508258 sameAs 2808508258 @default.
- W2808508258 citedByCount "1" @default.
- W2808508258 countsByYear W28085082582019 @default.
- W2808508258 crossrefType "posted-content" @default.
- W2808508258 hasAuthorship W2808508258A5022469843 @default.
- W2808508258 hasAuthorship W2808508258A5053299181 @default.
- W2808508258 hasAuthorship W2808508258A5063260720 @default.
- W2808508258 hasAuthorship W2808508258A5074723086 @default.
- W2808508258 hasConcept C101738243 @default.
- W2808508258 hasConcept C108583219 @default.
- W2808508258 hasConcept C111919701 @default.
- W2808508258 hasConcept C11413529 @default.
- W2808508258 hasConcept C119857082 @default.
- W2808508258 hasConcept C134306372 @default.
- W2808508258 hasConcept C154945302 @default.
- W2808508258 hasConcept C167966045 @default.
- W2808508258 hasConcept C182365436 @default.
- W2808508258 hasConcept C2780938662 @default.
- W2808508258 hasConcept C33923547 @default.
- W2808508258 hasConcept C39890363 @default.
- W2808508258 hasConcept C41008148 @default.
- W2808508258 hasConcept C51167844 @default.
- W2808508258 hasConcept C65965080 @default.
- W2808508258 hasConcept C77553402 @default.
- W2808508258 hasConceptScore W2808508258C101738243 @default.
- W2808508258 hasConceptScore W2808508258C108583219 @default.
- W2808508258 hasConceptScore W2808508258C111919701 @default.
- W2808508258 hasConceptScore W2808508258C11413529 @default.
- W2808508258 hasConceptScore W2808508258C119857082 @default.
- W2808508258 hasConceptScore W2808508258C134306372 @default.
- W2808508258 hasConceptScore W2808508258C154945302 @default.
- W2808508258 hasConceptScore W2808508258C167966045 @default.
- W2808508258 hasConceptScore W2808508258C182365436 @default.
- W2808508258 hasConceptScore W2808508258C2780938662 @default.
- W2808508258 hasConceptScore W2808508258C33923547 @default.
- W2808508258 hasConceptScore W2808508258C39890363 @default.
- W2808508258 hasConceptScore W2808508258C41008148 @default.
- W2808508258 hasConceptScore W2808508258C51167844 @default.
- W2808508258 hasConceptScore W2808508258C65965080 @default.
- W2808508258 hasConceptScore W2808508258C77553402 @default.
- W2808508258 hasLocation W28085082581 @default.
- W2808508258 hasOpenAccess W2808508258 @default.
- W2808508258 hasPrimaryLocation W28085082581 @default.
- W2808508258 hasRelatedWork W2012063769 @default.
- W2808508258 hasRelatedWork W2099057450 @default.
- W2808508258 hasRelatedWork W2114881090 @default.
- W2808508258 hasRelatedWork W2771375649 @default.
- W2808508258 hasRelatedWork W2788157105 @default.
- W2808508258 hasRelatedWork W2901490933 @default.
- W2808508258 hasRelatedWork W2912189848 @default.
- W2808508258 hasRelatedWork W2970966125 @default.
- W2808508258 hasRelatedWork W2971209830 @default.
- W2808508258 hasRelatedWork W2971970188 @default.
- W2808508258 hasRelatedWork W2972264839 @default.
- W2808508258 hasRelatedWork W2984061918 @default.
- W2808508258 hasRelatedWork W2996156127 @default.
- W2808508258 hasRelatedWork W3003780847 @default.
- W2808508258 hasRelatedWork W3035456141 @default.
- W2808508258 hasRelatedWork W3120211995 @default.
- W2808508258 hasRelatedWork W3158571811 @default.
- W2808508258 hasRelatedWork W3183912051 @default.
- W2808508258 hasRelatedWork W3197317199 @default.
- W2808508258 hasRelatedWork W2765540806 @default.
- W2808508258 isParatext "false" @default.
- W2808508258 isRetracted "false" @default.
- W2808508258 magId "2808508258" @default.
- W2808508258 workType "article" @default.