Matches in SemOpenAlex for { <https://semopenalex.org/work/W4323644089> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W4323644089 endingPage "685" @default.
- W4323644089 startingPage "670" @default.
- W4323644089 abstract "With the growth of neural network size, model compression has attracted increasing interest in recent research. As one of the most common techniques, pruning has been studied for a long time. By exploiting the structured sparsity of the neural network, existing methods can prune neurons instead of individual weights. However, in most existing pruning methods, surviving neurons are randomly connected in the neural network without any structure, and the non-zero weights within each neuron are also randomly distributed. Such irregular sparse structure can cause very high control overhead and irregular memory access for the hardware and even increase the neural network computational complexity. In this paper, we propose a three-layer hierarchical prior to promote a more regular sparse structure during pruning. The proposed three-layer hierarchical prior can achieve per-neuron weight-level structured sparsity and neuron-level structured sparsity. We derive an efficient Turbo-variational Bayesian inferencing (Turbo-VBI) algorithm to solve the resulting model compression problem with the proposed prior. The proposed Turbo-VBI algorithm has low complexity and can support more general priors than existing model compression algorithms. Simulation results show that our proposed algorithm can promote a more regular structure in the pruned neural networks while achieving even better performance in terms of compression rate and inferencing accuracy compared with the baselines." @default.
- W4323644089 created "2023-03-10" @default.
- W4323644089 creator A5040138091 @default.
- W4323644089 creator A5070509631 @default.
- W4323644089 creator A5091081870 @default.
- W4323644089 date "2023-01-01" @default.
- W4323644089 modified "2023-10-14" @default.
- W4323644089 title "Structured Bayesian Compression for Deep Neural Networks Based on the Turbo-VBI Approach" @default.
- W4323644089 cites W1540764732 @default.
- W4323644089 cites W2000721204 @default.
- W4323644089 cites W2154153158 @default.
- W4323644089 cites W2460144244 @default.
- W4323644089 cites W2554302513 @default.
- W4323644089 cites W2657126969 @default.
- W4323644089 cites W2948135617 @default.
- W4323644089 cites W2949866178 @default.
- W4323644089 cites W2982083293 @default.
- W4323644089 cites W2998470761 @default.
- W4323644089 cites W3006668132 @default.
- W4323644089 cites W3034288893 @default.
- W4323644089 cites W3043426275 @default.
- W4323644089 cites W4200089852 @default.
- W4323644089 cites W4288732494 @default.
- W4323644089 cites W4312796067 @default.
- W4323644089 doi "https://doi.org/10.1109/tsp.2023.3252165" @default.
- W4323644089 hasPublicationYear "2023" @default.
- W4323644089 type Work @default.
- W4323644089 citedByCount "0" @default.
- W4323644089 crossrefType "journal-article" @default.
- W4323644089 hasAuthorship W4323644089A5040138091 @default.
- W4323644089 hasAuthorship W4323644089A5070509631 @default.
- W4323644089 hasAuthorship W4323644089A5091081870 @default.
- W4323644089 hasBestOaLocation W43236440892 @default.
- W4323644089 hasConcept C108010975 @default.
- W4323644089 hasConcept C111919701 @default.
- W4323644089 hasConcept C11413529 @default.
- W4323644089 hasConcept C115961682 @default.
- W4323644089 hasConcept C127413603 @default.
- W4323644089 hasConcept C13481523 @default.
- W4323644089 hasConcept C154945302 @default.
- W4323644089 hasConcept C171146098 @default.
- W4323644089 hasConcept C179799912 @default.
- W4323644089 hasConcept C2776240298 @default.
- W4323644089 hasConcept C2779960059 @default.
- W4323644089 hasConcept C41008148 @default.
- W4323644089 hasConcept C50644808 @default.
- W4323644089 hasConcept C6557445 @default.
- W4323644089 hasConcept C86803240 @default.
- W4323644089 hasConcept C9417928 @default.
- W4323644089 hasConcept C94835093 @default.
- W4323644089 hasConceptScore W4323644089C108010975 @default.
- W4323644089 hasConceptScore W4323644089C111919701 @default.
- W4323644089 hasConceptScore W4323644089C11413529 @default.
- W4323644089 hasConceptScore W4323644089C115961682 @default.
- W4323644089 hasConceptScore W4323644089C127413603 @default.
- W4323644089 hasConceptScore W4323644089C13481523 @default.
- W4323644089 hasConceptScore W4323644089C154945302 @default.
- W4323644089 hasConceptScore W4323644089C171146098 @default.
- W4323644089 hasConceptScore W4323644089C179799912 @default.
- W4323644089 hasConceptScore W4323644089C2776240298 @default.
- W4323644089 hasConceptScore W4323644089C2779960059 @default.
- W4323644089 hasConceptScore W4323644089C41008148 @default.
- W4323644089 hasConceptScore W4323644089C50644808 @default.
- W4323644089 hasConceptScore W4323644089C6557445 @default.
- W4323644089 hasConceptScore W4323644089C86803240 @default.
- W4323644089 hasConceptScore W4323644089C9417928 @default.
- W4323644089 hasConceptScore W4323644089C94835093 @default.
- W4323644089 hasFunder F4320335480 @default.
- W4323644089 hasLocation W43236440891 @default.
- W4323644089 hasLocation W43236440892 @default.
- W4323644089 hasLocation W43236440893 @default.
- W4323644089 hasOpenAccess W4323644089 @default.
- W4323644089 hasPrimaryLocation W43236440891 @default.
- W4323644089 hasRelatedWork W1970992653 @default.
- W4323644089 hasRelatedWork W2030157244 @default.
- W4323644089 hasRelatedWork W2127738042 @default.
- W4323644089 hasRelatedWork W2132822875 @default.
- W4323644089 hasRelatedWork W2148412668 @default.
- W4323644089 hasRelatedWork W2171242668 @default.
- W4323644089 hasRelatedWork W2388867675 @default.
- W4323644089 hasRelatedWork W2785760655 @default.
- W4323644089 hasRelatedWork W2799771971 @default.
- W4323644089 hasRelatedWork W3189943615 @default.
- W4323644089 hasVolume "71" @default.
- W4323644089 isParatext "false" @default.
- W4323644089 isRetracted "false" @default.
- W4323644089 workType "article" @default.