FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W1762484328> ?p ?o ?g. }

• W1762484328 abstract "We describe the neural-network training framework used in the Kaldi speech recognition toolkit, which is geared towards training DNNs with large amounts of training data using multiple GPU-equipped or multicore machines. In order to be as hardwareagnostic as possible, we needed a way to use multiple machines without generating excessive network traffic. Our method is to average the neural network parameters periodically (typically every minute or two), and redistribute the averaged parameters to the machines for further training. Each machine sees different data. By itself, this method does not work very well. However, we have another method, an approximate and efficient implementation of Natural Gradient for Stochastic Gradient Descent (NG-SGD), which seems to allow our periodic-averaging method to work well, as well as substantially improving the convergence of SGD on a single machine." @default.
• W1762484328 created "2016-06-24" @default.
• W1762484328 creator A5001549192 @default.
• W1762484328 creator A5014580424 @default.
• W1762484328 creator A5084286453 @default.
• W1762484328 date "2014-01-01" @default.
• W1762484328 modified "2023-09-26" @default.
• W1762484328 title "Parallel training of Deep Neural Networks with Natural Gradient and Parameter Averaging" @default.
• W1762484328 cites W1499021337 @default.
• W1762484328 cites W1524333225 @default.
• W1762484328 cites W1533861849 @default.
• W1762484328 cites W1877570817 @default.
• W1762484328 cites W1970789124 @default.
• W1762484328 cites W1985371235 @default.
• W1762484328 cites W2000200144 @default.
• W1762484328 cites W2009320922 @default.
• W1762484328 cites W2026369565 @default.
• W1762484328 cites W2037740282 @default.
• W1762484328 cites W2104760318 @default.
• W1762484328 cites W2111284535 @default.
• W1762484328 cites W2131342762 @default.
• W1762484328 cites W2132211083 @default.
• W1762484328 cites W2136922672 @default.
• W1762484328 cites W2141778357 @default.
• W1762484328 cites W2148154194 @default.
• W1762484328 cites W2150769028 @default.
• W1762484328 cites W2150907703 @default.
• W1762484328 cites W2154887136 @default.
• W1762484328 cites W2160306971 @default.
• W1762484328 cites W2164273299 @default.
• W1762484328 cites W2167002981 @default.
• W1762484328 cites W2394932179 @default.
• W1762484328 cites W2914484425 @default.
• W1762484328 cites W2951781666 @default.
• W1762484328 cites W2962936867 @default.
• W1762484328 cites W38527073 @default.
• W1762484328 hasPublicationYear "2014" @default.
• W1762484328 type Work @default.
• W1762484328 sameAs 1762484328 @default.
• W1762484328 citedByCount "47" @default.
• W1762484328 countsByYear W17624843282015 @default.
• W1762484328 countsByYear W17624843282016 @default.
• W1762484328 countsByYear W17624843282017 @default.
• W1762484328 countsByYear W17624843282018 @default.
• W1762484328 countsByYear W17624843282019 @default.
• W1762484328 countsByYear W17624843282020 @default.
• W1762484328 countsByYear W17624843282021 @default.
• W1762484328 crossrefType "proceedings-article" @default.
• W1762484328 hasAuthorship W1762484328A5001549192 @default.
• W1762484328 hasAuthorship W1762484328A5014580424 @default.
• W1762484328 hasAuthorship W1762484328A5084286453 @default.
• W1762484328 hasConcept C119857082 @default.
• W1762484328 hasConcept C121332964 @default.
• W1762484328 hasConcept C127413603 @default.
• W1762484328 hasConcept C153258448 @default.
• W1762484328 hasConcept C153294291 @default.
• W1762484328 hasConcept C154945302 @default.
• W1762484328 hasConcept C162324750 @default.
• W1762484328 hasConcept C18762648 @default.
• W1762484328 hasConcept C206688291 @default.
• W1762484328 hasConcept C2777211547 @default.
• W1762484328 hasConcept C2777303404 @default.
• W1762484328 hasConcept C2984842247 @default.
• W1762484328 hasConcept C41008148 @default.
• W1762484328 hasConcept C50522688 @default.
• W1762484328 hasConcept C50644808 @default.
• W1762484328 hasConcept C78519656 @default.
• W1762484328 hasConceptScore W1762484328C119857082 @default.
• W1762484328 hasConceptScore W1762484328C121332964 @default.
• W1762484328 hasConceptScore W1762484328C127413603 @default.
• W1762484328 hasConceptScore W1762484328C153258448 @default.
• W1762484328 hasConceptScore W1762484328C153294291 @default.
• W1762484328 hasConceptScore W1762484328C154945302 @default.
• W1762484328 hasConceptScore W1762484328C162324750 @default.
• W1762484328 hasConceptScore W1762484328C18762648 @default.
• W1762484328 hasConceptScore W1762484328C206688291 @default.
• W1762484328 hasConceptScore W1762484328C2777211547 @default.
• W1762484328 hasConceptScore W1762484328C2777303404 @default.
• W1762484328 hasConceptScore W1762484328C2984842247 @default.
• W1762484328 hasConceptScore W1762484328C41008148 @default.
• W1762484328 hasConceptScore W1762484328C50522688 @default.
• W1762484328 hasConceptScore W1762484328C50644808 @default.
• W1762484328 hasConceptScore W1762484328C78519656 @default.
• W1762484328 hasLocation W17624843281 @default.
• W1762484328 hasOpenAccess W1762484328 @default.
• W1762484328 hasPrimaryLocation W17624843281 @default.
• W1762484328 hasRelatedWork W1494198834 @default.
• W1762484328 hasRelatedWork W1524333225 @default.
• W1762484328 hasRelatedWork W1821462560 @default.
• W1762484328 hasRelatedWork W2025198378 @default.
• W1762484328 hasRelatedWork W2026369565 @default.
• W1762484328 hasRelatedWork W2117671523 @default.
• W1762484328 hasRelatedWork W2131342762 @default.
• W1762484328 hasRelatedWork W2147768505 @default.
• W1762484328 hasRelatedWork W2150769028 @default.
• W1762484328 hasRelatedWork W2168231600 @default.
• W1762484328 hasRelatedWork W2293634267 @default.
• W1762484328 hasRelatedWork W2402146185 @default.
• W1762484328 hasRelatedWork W2405578611 @default.
• W1762484328 hasRelatedWork W2407022425 @default.

• next