Matches in SemOpenAlex for { <https://semopenalex.org/work/W4285024749> ?p ?o ?g. }
Showing items 1 to 90 of
90
with 100 items per page.
- W4285024749 abstract "Neural network models based on LSTM have been widely used in the fields of automatic speech recognition (ASR) and natural language processing (NLP). However, to obtain higher accuracy, the computational complexity and memory footprint of LSTM need to increase continually. In order to speed up LSTM inference, some previous works propose to compress the model with weight pruning. Row-Balanced Sparsity (RBS) is one of the excellent method because of balancing distribution of nonzero elements with negligible precision degradation. However, the compression ratio of the model is greatly limited due to its large index occupancy.A Share Index Row-Balanced Sparsity(SIRBS) compression method is presented in this paper, which shares the indices between rows in one row cluster. In this way, not only the nonzero weight distribution can be balanced with a very small cost of indices, but also decoding cost can be omitted. Compared with RBS, our overhead of indices can be reduced by 2x-8x. The performance and accuracy of our method is evaluated on the CPU using PyTorch-Kaldi toolkit. When the sparsity rate is 90%, compared with dense LSTM, our compression ratio and speedup are 6.7x-8.9x and 7.0x-9.9x respectively, and the accuracy only decreases by 0.2%-1.4%." @default.
- W4285024749 created "2022-07-12" @default.
- W4285024749 creator A5006683448 @default.
- W4285024749 creator A5019682335 @default.
- W4285024749 creator A5022974724 @default.
- W4285024749 creator A5024790920 @default.
- W4285024749 creator A5069310600 @default.
- W4285024749 creator A5088703563 @default.
- W4285024749 date "2022-03-01" @default.
- W4285024749 modified "2023-10-16" @default.
- W4285024749 title "Compressed LSTM using balanced sparsity" @default.
- W4285024749 doi "https://doi.org/10.1109/cbd54617.2021.00044" @default.
- W4285024749 hasPublicationYear "2022" @default.
- W4285024749 type Work @default.
- W4285024749 citedByCount "0" @default.
- W4285024749 crossrefType "proceedings-article" @default.
- W4285024749 hasAuthorship W4285024749A5006683448 @default.
- W4285024749 hasAuthorship W4285024749A5019682335 @default.
- W4285024749 hasAuthorship W4285024749A5022974724 @default.
- W4285024749 hasAuthorship W4285024749A5024790920 @default.
- W4285024749 hasAuthorship W4285024749A5069310600 @default.
- W4285024749 hasAuthorship W4285024749A5088703563 @default.
- W4285024749 hasConcept C108010975 @default.
- W4285024749 hasConcept C111919701 @default.
- W4285024749 hasConcept C11413529 @default.
- W4285024749 hasConcept C115961682 @default.
- W4285024749 hasConcept C127413603 @default.
- W4285024749 hasConcept C13481523 @default.
- W4285024749 hasConcept C137293760 @default.
- W4285024749 hasConcept C154945302 @default.
- W4285024749 hasConcept C171146098 @default.
- W4285024749 hasConcept C173608175 @default.
- W4285024749 hasConcept C2524010 @default.
- W4285024749 hasConcept C25797200 @default.
- W4285024749 hasConcept C2776214188 @default.
- W4285024749 hasConcept C2779960059 @default.
- W4285024749 hasConcept C33923547 @default.
- W4285024749 hasConcept C41008148 @default.
- W4285024749 hasConcept C50644808 @default.
- W4285024749 hasConcept C511840579 @default.
- W4285024749 hasConcept C57273362 @default.
- W4285024749 hasConcept C6557445 @default.
- W4285024749 hasConcept C68339613 @default.
- W4285024749 hasConcept C74912251 @default.
- W4285024749 hasConcept C86803240 @default.
- W4285024749 hasConcept C90805587 @default.
- W4285024749 hasConcept C9417928 @default.
- W4285024749 hasConcept C94835093 @default.
- W4285024749 hasConceptScore W4285024749C108010975 @default.
- W4285024749 hasConceptScore W4285024749C111919701 @default.
- W4285024749 hasConceptScore W4285024749C11413529 @default.
- W4285024749 hasConceptScore W4285024749C115961682 @default.
- W4285024749 hasConceptScore W4285024749C127413603 @default.
- W4285024749 hasConceptScore W4285024749C13481523 @default.
- W4285024749 hasConceptScore W4285024749C137293760 @default.
- W4285024749 hasConceptScore W4285024749C154945302 @default.
- W4285024749 hasConceptScore W4285024749C171146098 @default.
- W4285024749 hasConceptScore W4285024749C173608175 @default.
- W4285024749 hasConceptScore W4285024749C2524010 @default.
- W4285024749 hasConceptScore W4285024749C25797200 @default.
- W4285024749 hasConceptScore W4285024749C2776214188 @default.
- W4285024749 hasConceptScore W4285024749C2779960059 @default.
- W4285024749 hasConceptScore W4285024749C33923547 @default.
- W4285024749 hasConceptScore W4285024749C41008148 @default.
- W4285024749 hasConceptScore W4285024749C50644808 @default.
- W4285024749 hasConceptScore W4285024749C511840579 @default.
- W4285024749 hasConceptScore W4285024749C57273362 @default.
- W4285024749 hasConceptScore W4285024749C6557445 @default.
- W4285024749 hasConceptScore W4285024749C68339613 @default.
- W4285024749 hasConceptScore W4285024749C74912251 @default.
- W4285024749 hasConceptScore W4285024749C86803240 @default.
- W4285024749 hasConceptScore W4285024749C90805587 @default.
- W4285024749 hasConceptScore W4285024749C9417928 @default.
- W4285024749 hasConceptScore W4285024749C94835093 @default.
- W4285024749 hasLocation W42850247491 @default.
- W4285024749 hasOpenAccess W4285024749 @default.
- W4285024749 hasPrimaryLocation W42850247491 @default.
- W4285024749 hasRelatedWork W13657349 @default.
- W4285024749 hasRelatedWork W14230040 @default.
- W4285024749 hasRelatedWork W2433769 @default.
- W4285024749 hasRelatedWork W2494974 @default.
- W4285024749 hasRelatedWork W3836740 @default.
- W4285024749 hasRelatedWork W5568260 @default.
- W4285024749 hasRelatedWork W6143937 @default.
- W4285024749 hasRelatedWork W7155555 @default.
- W4285024749 hasRelatedWork W7232811 @default.
- W4285024749 hasRelatedWork W7401400 @default.
- W4285024749 isParatext "false" @default.
- W4285024749 isRetracted "false" @default.
- W4285024749 workType "article" @default.