FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2621265919> ?p ?o ?g. }

• W2621265919 abstract "Importance sampling has been successfully used to accelerate stochastic optimization in many convex problems. However, the lack of an efficient way to calculate the importance still hinders its application to Deep Learning. In this paper, we show that the loss value can be used as an alternative importance metric, and propose a way to efficiently approximate it for a deep model, using a small model trained for that purpose in parallel. This method allows in particular to utilize a biased gradient estimate that implicitly optimizes a soft max-loss, and leads to better generalization performance. While such method suffers from a prohibitively high variance of the gradient estimate when using a standard stochastic optimizer, we show that when it is combined with our sampling mechanism, it results in a reliable procedure. We showcase the generality of our method by testing it on both image classification and language modeling tasks using deep convolutional and recurrent neural networks. In particular, our method results in 30% faster training of a CNN for CIFAR10 than when using uniform sampling." @default.
• W2621265919 created "2017-06-09" @default.
• W2621265919 creator A5031829458 @default.
• W2621265919 creator A5076094010 @default.
• W2621265919 date "2017-05-31" @default.
• W2621265919 modified "2023-10-04" @default.
• W2621265919 title "Biased Importance Sampling for Deep Neural Network Training." @default.
• W2621265919 cites W1591801644 @default.
• W2621265919 cites W1632114991 @default.
• W2621265919 cites W1686810756 @default.
• W2621265919 cites W1842094663 @default.
• W2621265919 cites W2132984949 @default.
• W2621265919 cites W2135106139 @default.
• W2621265919 cites W2162287622 @default.
• W2621265919 cites W2164075197 @default.
• W2621265919 cites W2174940656 @default.
• W2621265919 cites W2177410802 @default.
• W2621265919 cites W2185726469 @default.
• W2621265919 cites W2271840356 @default.
• W2621265919 cites W2295869408 @default.
• W2621265919 cites W2296073425 @default.
• W2621265919 cites W2428862780 @default.
• W2621265919 cites W2964121744 @default.
• W2621265919 cites W3118608800 @default.
• W2621265919 cites W2298503502 @default.
• W2621265919 hasPublicationYear "2017" @default.
• W2621265919 type Work @default.
• W2621265919 sameAs 2621265919 @default.
• W2621265919 citedByCount "17" @default.
• W2621265919 countsByYear W26212659192018 @default.
• W2621265919 countsByYear W26212659192019 @default.
• W2621265919 countsByYear W26212659192020 @default.
• W2621265919 countsByYear W26212659192021 @default.
• W2621265919 crossrefType "posted-content" @default.
• W2621265919 hasAuthorship W2621265919A5031829458 @default.
• W2621265919 hasAuthorship W2621265919A5076094010 @default.
• W2621265919 hasConcept C108583219 @default.
• W2621265919 hasConcept C11413529 @default.
• W2621265919 hasConcept C115961682 @default.
• W2621265919 hasConcept C119857082 @default.
• W2621265919 hasConcept C121955636 @default.
• W2621265919 hasConcept C126255220 @default.
• W2621265919 hasConcept C134306372 @default.
• W2621265919 hasConcept C140779682 @default.
• W2621265919 hasConcept C144133560 @default.
• W2621265919 hasConcept C154945302 @default.
• W2621265919 hasConcept C15744967 @default.
• W2621265919 hasConcept C162324750 @default.
• W2621265919 hasConcept C176217482 @default.
• W2621265919 hasConcept C177148314 @default.
• W2621265919 hasConcept C196083921 @default.
• W2621265919 hasConcept C206688291 @default.
• W2621265919 hasConcept C21547014 @default.
• W2621265919 hasConcept C2780767217 @default.
• W2621265919 hasConcept C2984842247 @default.
• W2621265919 hasConcept C33923547 @default.
• W2621265919 hasConcept C41008148 @default.
• W2621265919 hasConcept C50644808 @default.
• W2621265919 hasConcept C542102704 @default.
• W2621265919 hasConcept C76155785 @default.
• W2621265919 hasConcept C81363708 @default.
• W2621265919 hasConcept C94915269 @default.
• W2621265919 hasConceptScore W2621265919C108583219 @default.
• W2621265919 hasConceptScore W2621265919C11413529 @default.
• W2621265919 hasConceptScore W2621265919C115961682 @default.
• W2621265919 hasConceptScore W2621265919C119857082 @default.
• W2621265919 hasConceptScore W2621265919C121955636 @default.
• W2621265919 hasConceptScore W2621265919C126255220 @default.
• W2621265919 hasConceptScore W2621265919C134306372 @default.
• W2621265919 hasConceptScore W2621265919C140779682 @default.
• W2621265919 hasConceptScore W2621265919C144133560 @default.
• W2621265919 hasConceptScore W2621265919C154945302 @default.
• W2621265919 hasConceptScore W2621265919C15744967 @default.
• W2621265919 hasConceptScore W2621265919C162324750 @default.
• W2621265919 hasConceptScore W2621265919C176217482 @default.
• W2621265919 hasConceptScore W2621265919C177148314 @default.
• W2621265919 hasConceptScore W2621265919C196083921 @default.
• W2621265919 hasConceptScore W2621265919C206688291 @default.
• W2621265919 hasConceptScore W2621265919C21547014 @default.
• W2621265919 hasConceptScore W2621265919C2780767217 @default.
• W2621265919 hasConceptScore W2621265919C2984842247 @default.
• W2621265919 hasConceptScore W2621265919C33923547 @default.
• W2621265919 hasConceptScore W2621265919C41008148 @default.
• W2621265919 hasConceptScore W2621265919C50644808 @default.
• W2621265919 hasConceptScore W2621265919C542102704 @default.
• W2621265919 hasConceptScore W2621265919C76155785 @default.
• W2621265919 hasConceptScore W2621265919C81363708 @default.
• W2621265919 hasConceptScore W2621265919C94915269 @default.
• W2621265919 hasLocation W26212659191 @default.
• W2621265919 hasOpenAccess W2621265919 @default.
• W2621265919 hasPrimaryLocation W26212659191 @default.
• W2621265919 hasRelatedWork W1842094663 @default.
• W2621265919 hasRelatedWork W2112796928 @default.
• W2621265919 hasRelatedWork W2132984949 @default.
• W2621265919 hasRelatedWork W2174940656 @default.
• W2621265919 hasRelatedWork W2177410802 @default.
• W2621265919 hasRelatedWork W2183341477 @default.
• W2621265919 hasRelatedWork W2194775991 @default.
• W2621265919 hasRelatedWork W2296073425 @default.
• W2621265919 hasRelatedWork W2335728318 @default.

• next