Matches in SemOpenAlex for { <https://semopenalex.org/work/W3170900164> ?p ?o ?g. }
- W3170900164 abstract "The use of artificial intelligence, Machine Learning and in particular Deep Learning (DL), have recently become a effective and standard de-facto solution for complex problems like image classification, sentiment analysis or natural language processing. In order to address the growing demand of performance of ML applications, research has focused on techniques for compressing the large amount of the parameters required by the Deep Neural Networks (DNN) used in DL. Some of these techniques include parameter pruning, weight-sharing, i.e. clustering of the weights, and parameter quantization. However, reducing the amount of parameters can lower the fault tolerance of DNNs, already sensitive to software and hardware faults caused by, among others, high particles strikes, row hammer or gradient descent attacks, et cetera. In this work we analyze the sensitivity to faults of widely used DNNs, in particular Convolutional Neural Networks (CNN), that have been compressed with the use of pruning, weight clustering and quantization. Our analysis shows that in DNNs that employ all such compression mechanisms, i.e. with their memory footprint reduced up to 86.3x, random single bit faults can result in accuracy drops up to 13.56%." @default.
- W3170900164 created "2021-06-22" @default.
- W3170900164 creator A5026569099 @default.
- W3170900164 creator A5043427541 @default.
- W3170900164 creator A5061274269 @default.
- W3170900164 creator A5087174287 @default.
- W3170900164 date "2021-01-01" @default.
- W3170900164 modified "2023-09-27" @default.
- W3170900164 title "Reliability Analysis of Compressed CNNs" @default.
- W3170900164 cites W1686810756 @default.
- W3170900164 cites W1895577753 @default.
- W3170900164 cites W1980782134 @default.
- W3170900164 cites W2112796928 @default.
- W3170900164 cites W2155428879 @default.
- W3170900164 cites W2155893237 @default.
- W3170900164 cites W2525778437 @default.
- W3170900164 cites W2762654171 @default.
- W3170900164 cites W2767260595 @default.
- W3170900164 cites W2771112233 @default.
- W3170900164 cites W2772825438 @default.
- W3170900164 cites W2809188712 @default.
- W3170900164 cites W2901848761 @default.
- W3170900164 cites W2903582334 @default.
- W3170900164 cites W2942091739 @default.
- W3170900164 cites W2943220429 @default.
- W3170900164 cites W2946206667 @default.
- W3170900164 cites W2964108906 @default.
- W3170900164 cites W2964299589 @default.
- W3170900164 cites W2981860227 @default.
- W3170900164 cites W2997106510 @default.
- W3170900164 cites W3041816437 @default.
- W3170900164 cites W3118608800 @default.
- W3170900164 hasPublicationYear "2021" @default.
- W3170900164 type Work @default.
- W3170900164 sameAs 3170900164 @default.
- W3170900164 citedByCount "0" @default.
- W3170900164 crossrefType "journal-article" @default.
- W3170900164 hasAuthorship W3170900164A5026569099 @default.
- W3170900164 hasAuthorship W3170900164A5043427541 @default.
- W3170900164 hasAuthorship W3170900164A5061274269 @default.
- W3170900164 hasAuthorship W3170900164A5087174287 @default.
- W3170900164 hasConcept C108010975 @default.
- W3170900164 hasConcept C108583219 @default.
- W3170900164 hasConcept C111919701 @default.
- W3170900164 hasConcept C11413529 @default.
- W3170900164 hasConcept C119857082 @default.
- W3170900164 hasConcept C120314980 @default.
- W3170900164 hasConcept C153180895 @default.
- W3170900164 hasConcept C154945302 @default.
- W3170900164 hasConcept C28855332 @default.
- W3170900164 hasConcept C41008148 @default.
- W3170900164 hasConcept C50644808 @default.
- W3170900164 hasConcept C63540848 @default.
- W3170900164 hasConcept C6557445 @default.
- W3170900164 hasConcept C73555534 @default.
- W3170900164 hasConcept C74912251 @default.
- W3170900164 hasConcept C81363708 @default.
- W3170900164 hasConcept C86803240 @default.
- W3170900164 hasConceptScore W3170900164C108010975 @default.
- W3170900164 hasConceptScore W3170900164C108583219 @default.
- W3170900164 hasConceptScore W3170900164C111919701 @default.
- W3170900164 hasConceptScore W3170900164C11413529 @default.
- W3170900164 hasConceptScore W3170900164C119857082 @default.
- W3170900164 hasConceptScore W3170900164C120314980 @default.
- W3170900164 hasConceptScore W3170900164C153180895 @default.
- W3170900164 hasConceptScore W3170900164C154945302 @default.
- W3170900164 hasConceptScore W3170900164C28855332 @default.
- W3170900164 hasConceptScore W3170900164C41008148 @default.
- W3170900164 hasConceptScore W3170900164C50644808 @default.
- W3170900164 hasConceptScore W3170900164C63540848 @default.
- W3170900164 hasConceptScore W3170900164C6557445 @default.
- W3170900164 hasConceptScore W3170900164C73555534 @default.
- W3170900164 hasConceptScore W3170900164C74912251 @default.
- W3170900164 hasConceptScore W3170900164C81363708 @default.
- W3170900164 hasConceptScore W3170900164C86803240 @default.
- W3170900164 hasLocation W31709001641 @default.
- W3170900164 hasOpenAccess W3170900164 @default.
- W3170900164 hasPrimaryLocation W31709001641 @default.
- W3170900164 hasRelatedWork W1724438581 @default.
- W3170900164 hasRelatedWork W2177847924 @default.
- W3170900164 hasRelatedWork W2575345876 @default.
- W3170900164 hasRelatedWork W2781820017 @default.
- W3170900164 hasRelatedWork W2798400493 @default.
- W3170900164 hasRelatedWork W2899742314 @default.
- W3170900164 hasRelatedWork W2903711666 @default.
- W3170900164 hasRelatedWork W2921388130 @default.
- W3170900164 hasRelatedWork W2952265507 @default.
- W3170900164 hasRelatedWork W3003530529 @default.
- W3170900164 hasRelatedWork W3006243129 @default.
- W3170900164 hasRelatedWork W3035826382 @default.
- W3170900164 hasRelatedWork W3117674360 @default.
- W3170900164 hasRelatedWork W3119853956 @default.
- W3170900164 hasRelatedWork W3131103899 @default.
- W3170900164 hasRelatedWork W3176213365 @default.
- W3170900164 hasRelatedWork W3179748517 @default.
- W3170900164 hasRelatedWork W3195381335 @default.
- W3170900164 hasRelatedWork W3202728214 @default.
- W3170900164 hasRelatedWork W3214436459 @default.
- W3170900164 isParatext "false" @default.
- W3170900164 isRetracted "false" @default.