Matches in SemOpenAlex for { <https://semopenalex.org/work/W2951985269> ?p ?o ?g. }
- W2951985269 abstract "Choosing appropriate architectures and regularization strategies for deep networks is crucial to good predictive performance. To shed light on this problem, we analyze the analogous problem of constructing useful priors on compositions of functions. Specifically, we study the deep Gaussian process, a type of infinitely-wide, deep neural network. We show that in standard architectures, the representational capacity of the network tends to capture fewer degrees of freedom as the number of layers increases, retaining only a single degree of freedom in the limit. We propose an alternate network architecture which does not suffer from this pathology. We also examine deep covariance functions, obtained by composing infinitely many feature transforms. Lastly, we characterize the class of models obtained by performing dropout on Gaussian processes." @default.
- W2951985269 created "2019-06-27" @default.
- W2951985269 creator A5007769527 @default.
- W2951985269 creator A5010820865 @default.
- W2951985269 creator A5022938074 @default.
- W2951985269 creator A5030409494 @default.
- W2951985269 date "2014-02-24" @default.
- W2951985269 modified "2023-09-27" @default.
- W2951985269 title "Avoiding pathologies in very deep networks" @default.
- W2951985269 cites W1567512734 @default.
- W2951985269 cites W16016350 @default.
- W2951985269 cites W1641973859 @default.
- W2951985269 cites W1904365287 @default.
- W2951985269 cites W196761320 @default.
- W2951985269 cites W2098626000 @default.
- W2951985269 cites W2115781030 @default.
- W2951985269 cites W2125569215 @default.
- W2951985269 cites W2130283669 @default.
- W2951985269 cites W2133257461 @default.
- W2951985269 cites W2145439642 @default.
- W2951985269 cites W2145860534 @default.
- W2951985269 cites W2159687189 @default.
- W2951985269 cites W2162724919 @default.
- W2951985269 cites W2167608136 @default.
- W2951985269 cites W2183112036 @default.
- W2951985269 cites W2218318129 @default.
- W2951985269 cites W2294567968 @default.
- W2951985269 cites W2949190276 @default.
- W2951985269 cites W2951308445 @default.
- W2951985269 cites W2952278564 @default.
- W2951985269 cites W2963711523 @default.
- W2951985269 cites W35527955 @default.
- W2951985269 cites W78356000 @default.
- W2951985269 cites W2066350599 @default.
- W2951985269 hasPublicationYear "2014" @default.
- W2951985269 type Work @default.
- W2951985269 sameAs 2951985269 @default.
- W2951985269 citedByCount "22" @default.
- W2951985269 countsByYear W29519852692014 @default.
- W2951985269 countsByYear W29519852692015 @default.
- W2951985269 countsByYear W29519852692016 @default.
- W2951985269 countsByYear W29519852692017 @default.
- W2951985269 countsByYear W29519852692018 @default.
- W2951985269 countsByYear W29519852692020 @default.
- W2951985269 countsByYear W29519852692021 @default.
- W2951985269 crossrefType "posted-content" @default.
- W2951985269 hasAuthorship W2951985269A5007769527 @default.
- W2951985269 hasAuthorship W2951985269A5010820865 @default.
- W2951985269 hasAuthorship W2951985269A5022938074 @default.
- W2951985269 hasAuthorship W2951985269A5030409494 @default.
- W2951985269 hasConcept C105795698 @default.
- W2951985269 hasConcept C107673813 @default.
- W2951985269 hasConcept C108583219 @default.
- W2951985269 hasConcept C11413529 @default.
- W2951985269 hasConcept C119857082 @default.
- W2951985269 hasConcept C121332964 @default.
- W2951985269 hasConcept C134306372 @default.
- W2951985269 hasConcept C151201525 @default.
- W2951985269 hasConcept C154945302 @default.
- W2951985269 hasConcept C163716315 @default.
- W2951985269 hasConcept C177769412 @default.
- W2951985269 hasConcept C178650346 @default.
- W2951985269 hasConcept C193415008 @default.
- W2951985269 hasConcept C2776135515 @default.
- W2951985269 hasConcept C2776145597 @default.
- W2951985269 hasConcept C2984842247 @default.
- W2951985269 hasConcept C33923547 @default.
- W2951985269 hasConcept C38652104 @default.
- W2951985269 hasConcept C41008148 @default.
- W2951985269 hasConcept C61326573 @default.
- W2951985269 hasConcept C62520636 @default.
- W2951985269 hasConceptScore W2951985269C105795698 @default.
- W2951985269 hasConceptScore W2951985269C107673813 @default.
- W2951985269 hasConceptScore W2951985269C108583219 @default.
- W2951985269 hasConceptScore W2951985269C11413529 @default.
- W2951985269 hasConceptScore W2951985269C119857082 @default.
- W2951985269 hasConceptScore W2951985269C121332964 @default.
- W2951985269 hasConceptScore W2951985269C134306372 @default.
- W2951985269 hasConceptScore W2951985269C151201525 @default.
- W2951985269 hasConceptScore W2951985269C154945302 @default.
- W2951985269 hasConceptScore W2951985269C163716315 @default.
- W2951985269 hasConceptScore W2951985269C177769412 @default.
- W2951985269 hasConceptScore W2951985269C178650346 @default.
- W2951985269 hasConceptScore W2951985269C193415008 @default.
- W2951985269 hasConceptScore W2951985269C2776135515 @default.
- W2951985269 hasConceptScore W2951985269C2776145597 @default.
- W2951985269 hasConceptScore W2951985269C2984842247 @default.
- W2951985269 hasConceptScore W2951985269C33923547 @default.
- W2951985269 hasConceptScore W2951985269C38652104 @default.
- W2951985269 hasConceptScore W2951985269C41008148 @default.
- W2951985269 hasConceptScore W2951985269C61326573 @default.
- W2951985269 hasConceptScore W2951985269C62520636 @default.
- W2951985269 hasLocation W29519852691 @default.
- W2951985269 hasOpenAccess W2951985269 @default.
- W2951985269 hasPrimaryLocation W29519852691 @default.
- W2951985269 hasRelatedWork W137285897 @default.
- W2951985269 hasRelatedWork W1567512734 @default.
- W2951985269 hasRelatedWork W1571870753 @default.
- W2951985269 hasRelatedWork W1746819321 @default.
- W2951985269 hasRelatedWork W1909320841 @default.