Matches in SemOpenAlex for { <https://semopenalex.org/work/W2785334672> ?p ?o ?g. }
Showing items 1 to 81 of
81
with 100 items per page.
- W2785334672 abstract "In this work we explore a straightforward variational Bayes scheme for Recurrent Neural Networks. Firstly, we show that a simple adaptation of truncated backpropagation through time can yield good quality uncertainty estimates and superior regularisation at only a small extra computational cost during training, also reducing the amount of parameters by 80%. Secondly, we demonstrate how a novel kind of posterior approximation yields further improvements to the performance of Bayesian RNNs. We incorporate local gradient information into the approximate posterior to sharpen it around the current batch statistics. We show how this technique is not exclusive to recurrent neural networks and can be applied more widely to train Bayesian neural networks. We also empirically demonstrate how Bayesian RNNs are superior to traditional RNNs on a language modelling benchmark and an image captioning task, as well as showing how each of these methods improve our model over a variety of other schemes for training them. We also introduce a new benchmark for studying uncertainty for language models so future methods can be easily compared." @default.
- W2785334672 created "2018-02-23" @default.
- W2785334672 creator A5003562101 @default.
- W2785334672 creator A5030688126 @default.
- W2785334672 creator A5088815712 @default.
- W2785334672 date "2018-02-15" @default.
- W2785334672 modified "2023-09-27" @default.
- W2785334672 title "Revisiting Bayes by Backprop" @default.
- W2785334672 hasPublicationYear "2018" @default.
- W2785334672 type Work @default.
- W2785334672 sameAs 2785334672 @default.
- W2785334672 citedByCount "0" @default.
- W2785334672 crossrefType "journal-article" @default.
- W2785334672 hasAuthorship W2785334672A5003562101 @default.
- W2785334672 hasAuthorship W2785334672A5030688126 @default.
- W2785334672 hasAuthorship W2785334672A5088815712 @default.
- W2785334672 hasConcept C107673813 @default.
- W2785334672 hasConcept C115961682 @default.
- W2785334672 hasConcept C119857082 @default.
- W2785334672 hasConcept C13280743 @default.
- W2785334672 hasConcept C136197465 @default.
- W2785334672 hasConcept C137293760 @default.
- W2785334672 hasConcept C147168706 @default.
- W2785334672 hasConcept C154945302 @default.
- W2785334672 hasConcept C155032097 @default.
- W2785334672 hasConcept C157657479 @default.
- W2785334672 hasConcept C162324750 @default.
- W2785334672 hasConcept C185798385 @default.
- W2785334672 hasConcept C187736073 @default.
- W2785334672 hasConcept C205649164 @default.
- W2785334672 hasConcept C207201462 @default.
- W2785334672 hasConcept C2780451532 @default.
- W2785334672 hasConcept C33724603 @default.
- W2785334672 hasConcept C41008148 @default.
- W2785334672 hasConcept C50644808 @default.
- W2785334672 hasConceptScore W2785334672C107673813 @default.
- W2785334672 hasConceptScore W2785334672C115961682 @default.
- W2785334672 hasConceptScore W2785334672C119857082 @default.
- W2785334672 hasConceptScore W2785334672C13280743 @default.
- W2785334672 hasConceptScore W2785334672C136197465 @default.
- W2785334672 hasConceptScore W2785334672C137293760 @default.
- W2785334672 hasConceptScore W2785334672C147168706 @default.
- W2785334672 hasConceptScore W2785334672C154945302 @default.
- W2785334672 hasConceptScore W2785334672C155032097 @default.
- W2785334672 hasConceptScore W2785334672C157657479 @default.
- W2785334672 hasConceptScore W2785334672C162324750 @default.
- W2785334672 hasConceptScore W2785334672C185798385 @default.
- W2785334672 hasConceptScore W2785334672C187736073 @default.
- W2785334672 hasConceptScore W2785334672C205649164 @default.
- W2785334672 hasConceptScore W2785334672C207201462 @default.
- W2785334672 hasConceptScore W2785334672C2780451532 @default.
- W2785334672 hasConceptScore W2785334672C33724603 @default.
- W2785334672 hasConceptScore W2785334672C41008148 @default.
- W2785334672 hasConceptScore W2785334672C50644808 @default.
- W2785334672 hasLocation W27853346721 @default.
- W2785334672 hasOpenAccess W2785334672 @default.
- W2785334672 hasPrimaryLocation W27853346721 @default.
- W2785334672 hasRelatedWork W142921608 @default.
- W2785334672 hasRelatedWork W2269264513 @default.
- W2785334672 hasRelatedWork W2437421599 @default.
- W2785334672 hasRelatedWork W2556372419 @default.
- W2785334672 hasRelatedWork W2781494504 @default.
- W2785334672 hasRelatedWork W2808181858 @default.
- W2785334672 hasRelatedWork W2897001865 @default.
- W2785334672 hasRelatedWork W2899648426 @default.
- W2785334672 hasRelatedWork W2910171846 @default.
- W2785334672 hasRelatedWork W2950117601 @default.
- W2785334672 hasRelatedWork W2950323007 @default.
- W2785334672 hasRelatedWork W2963859455 @default.
- W2785334672 hasRelatedWork W2964711111 @default.
- W2785334672 hasRelatedWork W2978400056 @default.
- W2785334672 hasRelatedWork W3016451125 @default.
- W2785334672 hasRelatedWork W3037722013 @default.
- W2785334672 hasRelatedWork W3064582752 @default.
- W2785334672 hasRelatedWork W3130438476 @default.
- W2785334672 hasRelatedWork W3192802934 @default.
- W2785334672 hasRelatedWork W601603264 @default.
- W2785334672 isParatext "false" @default.
- W2785334672 isRetracted "false" @default.
- W2785334672 magId "2785334672" @default.
- W2785334672 workType "article" @default.