Matches in SemOpenAlex for { <https://semopenalex.org/work/W2808483802> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W2808483802 abstract "Deep neural networks (DNNs) have achieved great success, but the applications to mobile devices are limited due to their huge model size and low inference speed. Much effort thus has been devoted to pruning DNNs. Layer-wise neuron pruning methods have shown their effectiveness, which minimize the reconstruction error of linear response with a limited number of neurons in each single layer pruning. In this paper, we propose a new layer-wise neuron pruning approach by minimizing the reconstruction error of nonlinear units, which might be more reasonable since the error before and after activation can change significantly. An iterative optimization procedure combining greedy selection with gradient decent is proposed for single layer pruning. Experimental results on benchmark DNN models show the superiority of the proposed approach. Particularly, for VGGNet, the proposed approach can compress its disk space by 13.6× and bring a speedup of 3.7×; for AlexNet, it can achieve a compression rate of 4.1× and a speedup of 2.2×, respectively." @default.
- W2808483802 created "2018-06-21" @default.
- W2808483802 creator A5001344682 @default.
- W2808483802 creator A5014258107 @default.
- W2808483802 creator A5021254405 @default.
- W2808483802 creator A5064249897 @default.
- W2808483802 date "2018-07-01" @default.
- W2808483802 modified "2023-09-23" @default.
- W2808483802 title "Efficient DNN Neuron Pruning by Minimizing Layer-wise Nonlinear Reconstruction Error" @default.
- W2808483802 doi "https://doi.org/10.24963/ijcai.2018/318" @default.
- W2808483802 hasPublicationYear "2018" @default.
- W2808483802 type Work @default.
- W2808483802 sameAs 2808483802 @default.
- W2808483802 citedByCount "27" @default.
- W2808483802 countsByYear W28084838022018 @default.
- W2808483802 countsByYear W28084838022019 @default.
- W2808483802 countsByYear W28084838022020 @default.
- W2808483802 countsByYear W28084838022021 @default.
- W2808483802 countsByYear W28084838022022 @default.
- W2808483802 countsByYear W28084838022023 @default.
- W2808483802 crossrefType "proceedings-article" @default.
- W2808483802 hasAuthorship W2808483802A5001344682 @default.
- W2808483802 hasAuthorship W2808483802A5014258107 @default.
- W2808483802 hasAuthorship W2808483802A5021254405 @default.
- W2808483802 hasAuthorship W2808483802A5064249897 @default.
- W2808483802 hasBestOaLocation W28084838021 @default.
- W2808483802 hasConcept C108010975 @default.
- W2808483802 hasConcept C11413529 @default.
- W2808483802 hasConcept C121332964 @default.
- W2808483802 hasConcept C13280743 @default.
- W2808483802 hasConcept C154945302 @default.
- W2808483802 hasConcept C158622935 @default.
- W2808483802 hasConcept C173608175 @default.
- W2808483802 hasConcept C178790620 @default.
- W2808483802 hasConcept C185592680 @default.
- W2808483802 hasConcept C185798385 @default.
- W2808483802 hasConcept C205649164 @default.
- W2808483802 hasConcept C2776214188 @default.
- W2808483802 hasConcept C2779227376 @default.
- W2808483802 hasConcept C40969351 @default.
- W2808483802 hasConcept C41008148 @default.
- W2808483802 hasConcept C50644808 @default.
- W2808483802 hasConcept C62520636 @default.
- W2808483802 hasConcept C6557445 @default.
- W2808483802 hasConcept C68339613 @default.
- W2808483802 hasConcept C86803240 @default.
- W2808483802 hasConceptScore W2808483802C108010975 @default.
- W2808483802 hasConceptScore W2808483802C11413529 @default.
- W2808483802 hasConceptScore W2808483802C121332964 @default.
- W2808483802 hasConceptScore W2808483802C13280743 @default.
- W2808483802 hasConceptScore W2808483802C154945302 @default.
- W2808483802 hasConceptScore W2808483802C158622935 @default.
- W2808483802 hasConceptScore W2808483802C173608175 @default.
- W2808483802 hasConceptScore W2808483802C178790620 @default.
- W2808483802 hasConceptScore W2808483802C185592680 @default.
- W2808483802 hasConceptScore W2808483802C185798385 @default.
- W2808483802 hasConceptScore W2808483802C205649164 @default.
- W2808483802 hasConceptScore W2808483802C2776214188 @default.
- W2808483802 hasConceptScore W2808483802C2779227376 @default.
- W2808483802 hasConceptScore W2808483802C40969351 @default.
- W2808483802 hasConceptScore W2808483802C41008148 @default.
- W2808483802 hasConceptScore W2808483802C50644808 @default.
- W2808483802 hasConceptScore W2808483802C62520636 @default.
- W2808483802 hasConceptScore W2808483802C6557445 @default.
- W2808483802 hasConceptScore W2808483802C68339613 @default.
- W2808483802 hasConceptScore W2808483802C86803240 @default.
- W2808483802 hasLocation W28084838021 @default.
- W2808483802 hasOpenAccess W2808483802 @default.
- W2808483802 hasPrimaryLocation W28084838021 @default.
- W2808483802 hasRelatedWork W2053732522 @default.
- W2808483802 hasRelatedWork W2348739446 @default.
- W2808483802 hasRelatedWork W2899244816 @default.
- W2808483802 hasRelatedWork W3011895793 @default.
- W2808483802 hasRelatedWork W3088334494 @default.
- W2808483802 hasRelatedWork W3119997217 @default.
- W2808483802 hasRelatedWork W3173602248 @default.
- W2808483802 hasRelatedWork W3196410085 @default.
- W2808483802 hasRelatedWork W3207072508 @default.
- W2808483802 hasRelatedWork W4323547442 @default.
- W2808483802 isParatext "false" @default.
- W2808483802 isRetracted "false" @default.
- W2808483802 magId "2808483802" @default.
- W2808483802 workType "article" @default.