FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2948634622> ?p ?o ?g. }

• W2948634622 abstract "In this paper we present a novel and general method to accelerate convolutional neural network (CNN) inference by taking advantage of feature map sparsity. We experimentally demonstrate that a highly quantized version of the original network is sufficient in predicting the output sparsity accurately, and verify that leveraging such sparsity in inference incurs negligible accuracy drop compared with the original network. To accelerate inference, for each convolution layer our approach first obtains a binary sparsity mask of the output feature maps by running inference on a quantized version of the original network layer, and then conducts a full-precision sparse convolution to find out the precise values of the non-zero outputs. Compared with existing work, our approach avoids the overhead of training additional auxiliary networks, while is still applicable to general CNN networks without being limited to certain application domains." @default.
• W2948634622 created "2019-06-14" @default.
• W2948634622 creator A5000260708 @default.
• W2948634622 creator A5021104158 @default.
• W2948634622 creator A5026864081 @default.
• W2948634622 creator A5027782298 @default.
• W2948634622 creator A5034855502 @default.
• W2948634622 creator A5061764261 @default.
• W2948634622 creator A5085795040 @default.
• W2948634622 creator A5086945155 @default.
• W2948634622 date "2019-06-01" @default.
• W2948634622 modified "2023-10-06" @default.
• W2948634622 title "SeerNet: Predicting Convolutional Neural Network Feature-Map Sparsity Through Low-Bit Quantization" @default.
• W2948634622 cites W1996901117 @default.
• W2948634622 cites W2097117768 @default.
• W2948634622 cites W2117539524 @default.
• W2948634622 cites W2183341477 @default.
• W2948634622 cites W2194775991 @default.
• W2948634622 cites W2400429454 @default.
• W2948634622 cites W2562731582 @default.
• W2948634622 cites W2585720638 @default.
• W2948634622 cites W2604998962 @default.
• W2948634622 cites W2606492274 @default.
• W2948634622 cites W2625457103 @default.
• W2948634622 cites W2883542588 @default.
• W2948634622 cites W2884071170 @default.
• W2948634622 cites W2897529137 @default.
• W2948634622 cites W2915106038 @default.
• W2948634622 cites W2962917547 @default.
• W2948634622 cites W2963521187 @default.
• W2948634622 cites W2963896595 @default.
• W2948634622 doi "https://doi.org/10.1109/cvpr.2019.01147" @default.
• W2948634622 hasPublicationYear "2019" @default.
• W2948634622 type Work @default.
• W2948634622 sameAs 2948634622 @default.
• W2948634622 citedByCount "55" @default.
• W2948634622 countsByYear W29486346222019 @default.
• W2948634622 countsByYear W29486346222020 @default.
• W2948634622 countsByYear W29486346222021 @default.
• W2948634622 countsByYear W29486346222022 @default.
• W2948634622 countsByYear W29486346222023 @default.
• W2948634622 crossrefType "proceedings-article" @default.
• W2948634622 hasAuthorship W2948634622A5000260708 @default.
• W2948634622 hasAuthorship W2948634622A5021104158 @default.
• W2948634622 hasAuthorship W2948634622A5026864081 @default.
• W2948634622 hasAuthorship W2948634622A5027782298 @default.
• W2948634622 hasAuthorship W2948634622A5034855502 @default.
• W2948634622 hasAuthorship W2948634622A5061764261 @default.
• W2948634622 hasAuthorship W2948634622A5085795040 @default.
• W2948634622 hasAuthorship W2948634622A5086945155 @default.
• W2948634622 hasConcept C111919701 @default.
• W2948634622 hasConcept C11413529 @default.
• W2948634622 hasConcept C138885662 @default.
• W2948634622 hasConcept C153180895 @default.
• W2948634622 hasConcept C154945302 @default.
• W2948634622 hasConcept C2776214188 @default.
• W2948634622 hasConcept C2776401178 @default.
• W2948634622 hasConcept C2779960059 @default.
• W2948634622 hasConcept C28855332 @default.
• W2948634622 hasConcept C33923547 @default.
• W2948634622 hasConcept C41008148 @default.
• W2948634622 hasConcept C41895202 @default.
• W2948634622 hasConcept C45347329 @default.
• W2948634622 hasConcept C48372109 @default.
• W2948634622 hasConcept C50644808 @default.
• W2948634622 hasConcept C81363708 @default.
• W2948634622 hasConcept C94375191 @default.
• W2948634622 hasConceptScore W2948634622C111919701 @default.
• W2948634622 hasConceptScore W2948634622C11413529 @default.
• W2948634622 hasConceptScore W2948634622C138885662 @default.
• W2948634622 hasConceptScore W2948634622C153180895 @default.
• W2948634622 hasConceptScore W2948634622C154945302 @default.
• W2948634622 hasConceptScore W2948634622C2776214188 @default.
• W2948634622 hasConceptScore W2948634622C2776401178 @default.
• W2948634622 hasConceptScore W2948634622C2779960059 @default.
• W2948634622 hasConceptScore W2948634622C28855332 @default.
• W2948634622 hasConceptScore W2948634622C33923547 @default.
• W2948634622 hasConceptScore W2948634622C41008148 @default.
• W2948634622 hasConceptScore W2948634622C41895202 @default.
• W2948634622 hasConceptScore W2948634622C45347329 @default.
• W2948634622 hasConceptScore W2948634622C48372109 @default.
• W2948634622 hasConceptScore W2948634622C50644808 @default.
• W2948634622 hasConceptScore W2948634622C81363708 @default.
• W2948634622 hasConceptScore W2948634622C94375191 @default.
• W2948634622 hasLocation W29486346221 @default.
• W2948634622 hasOpenAccess W2948634622 @default.
• W2948634622 hasPrimaryLocation W29486346221 @default.
• W2948634622 hasRelatedWork W1728708896 @default.
• W2948634622 hasRelatedWork W2295021132 @default.
• W2948634622 hasRelatedWork W2572787276 @default.
• W2948634622 hasRelatedWork W2737652487 @default.
• W2948634622 hasRelatedWork W2760085659 @default.
• W2948634622 hasRelatedWork W2886673456 @default.
• W2948634622 hasRelatedWork W3026025619 @default.
• W2948634622 hasRelatedWork W3106036237 @default.
• W2948634622 hasRelatedWork W4206841102 @default.
• W2948634622 hasRelatedWork W4225145356 @default.
• W2948634622 isParatext "false" @default.
• W2948634622 isRetracted "false" @default.
• W2948634622 magId "2948634622" @default.

• next