Matches in SemOpenAlex for { <https://semopenalex.org/work/W3003896899> ?p ?o ?g. }
- W3003896899 abstract "Quantization neural networks (QNNs) are very attractive to the industry because their extremely cheap calculation and storage overhead, but their performance is still worse than that of networks with full-precision parameters. Most of existing methods aim to enhance performance of QNNs especially binary neural networks by exploiting more effective training techniques. However, we find the representation capability of quantization features is far weaker than full-precision features by experiments. We address this problem by projecting features in original full-precision networks to high-dimensional quantization features. Simultaneously, redundant quantization features will be eliminated in order to avoid unrestricted growth of dimensions for some datasets. Then, a compact quantization neural network but with sufficient representation ability will be established. Experimental results on benchmark datasets demonstrate that the proposed method is able to establish QNNs with much less parameters and calculations but almost the same performance as that of full-precision baseline models, e.g. $29.9%$ top-1 error of binary ResNet-18 on the ImageNet ILSVRC 2012 dataset." @default.
- W3003896899 created "2020-02-07" @default.
- W3003896899 creator A5031128205 @default.
- W3003896899 creator A5047455588 @default.
- W3003896899 creator A5056485902 @default.
- W3003896899 creator A5062581018 @default.
- W3003896899 creator A5066602150 @default.
- W3003896899 creator A5081216284 @default.
- W3003896899 date "2020-02-02" @default.
- W3003896899 modified "2023-10-14" @default.
- W3003896899 title "Widening and Squeezing: Towards Accurate and Efficient QNNs" @default.
- W3003896899 cites W138476454 @default.
- W3003896899 cites W1821462560 @default.
- W3003896899 cites W1903029394 @default.
- W3003896899 cites W2001141328 @default.
- W3003896899 cites W2013305145 @default.
- W3003896899 cites W2053186076 @default.
- W3003896899 cites W2097117768 @default.
- W3003896899 cites W2102605133 @default.
- W3003896899 cites W2112796928 @default.
- W3003896899 cites W2117539524 @default.
- W3003896899 cites W2120972216 @default.
- W3003896899 cites W2125389748 @default.
- W3003896899 cites W2154872931 @default.
- W3003896899 cites W2163605009 @default.
- W3003896899 cites W2194775991 @default.
- W3003896899 cites W2279098554 @default.
- W3003896899 cites W2300242332 @default.
- W3003896899 cites W2319920447 @default.
- W3003896899 cites W2469490737 @default.
- W3003896899 cites W2612445135 @default.
- W3003896899 cites W2613718673 @default.
- W3003896899 cites W2808168148 @default.
- W3003896899 cites W2884150179 @default.
- W3003896899 cites W2887447938 @default.
- W3003896899 cites W2908314593 @default.
- W3003896899 cites W2928560789 @default.
- W3003896899 cites W2949829435 @default.
- W3003896899 cites W2962835968 @default.
- W3003896899 cites W2962851801 @default.
- W3003896899 cites W2962965870 @default.
- W3003896899 cites W2962988160 @default.
- W3003896899 cites W2963048316 @default.
- W3003896899 cites W2963114950 @default.
- W3003896899 cites W2963125010 @default.
- W3003896899 cites W2963363373 @default.
- W3003896899 cites W2963674932 @default.
- W3003896899 cites W2964008506 @default.
- W3003896899 cites W2964118293 @default.
- W3003896899 cites W2964288706 @default.
- W3003896899 cites W2964299589 @default.
- W3003896899 cites W2964622438 @default.
- W3003896899 cites W2970691381 @default.
- W3003896899 cites W2982234100 @default.
- W3003896899 cites W3106250896 @default.
- W3003896899 cites W3118608800 @default.
- W3003896899 cites W587794757 @default.
- W3003896899 doi "https://doi.org/10.48550/arxiv.2002.00555" @default.
- W3003896899 hasPublicationYear "2020" @default.
- W3003896899 type Work @default.
- W3003896899 sameAs 3003896899 @default.
- W3003896899 citedByCount "1" @default.
- W3003896899 countsByYear W30038968992020 @default.
- W3003896899 crossrefType "posted-content" @default.
- W3003896899 hasAuthorship W3003896899A5031128205 @default.
- W3003896899 hasAuthorship W3003896899A5047455588 @default.
- W3003896899 hasAuthorship W3003896899A5056485902 @default.
- W3003896899 hasAuthorship W3003896899A5062581018 @default.
- W3003896899 hasAuthorship W3003896899A5066602150 @default.
- W3003896899 hasAuthorship W3003896899A5081216284 @default.
- W3003896899 hasBestOaLocation W30038968991 @default.
- W3003896899 hasConcept C11413529 @default.
- W3003896899 hasConcept C119857082 @default.
- W3003896899 hasConcept C13280743 @default.
- W3003896899 hasConcept C153180895 @default.
- W3003896899 hasConcept C154945302 @default.
- W3003896899 hasConcept C17744445 @default.
- W3003896899 hasConcept C185798385 @default.
- W3003896899 hasConcept C199539241 @default.
- W3003896899 hasConcept C205649164 @default.
- W3003896899 hasConcept C2776359362 @default.
- W3003896899 hasConcept C28855332 @default.
- W3003896899 hasConcept C2984842247 @default.
- W3003896899 hasConcept C33923547 @default.
- W3003896899 hasConcept C41008148 @default.
- W3003896899 hasConcept C48372109 @default.
- W3003896899 hasConcept C50644808 @default.
- W3003896899 hasConcept C94375191 @default.
- W3003896899 hasConcept C94625758 @default.
- W3003896899 hasConceptScore W3003896899C11413529 @default.
- W3003896899 hasConceptScore W3003896899C119857082 @default.
- W3003896899 hasConceptScore W3003896899C13280743 @default.
- W3003896899 hasConceptScore W3003896899C153180895 @default.
- W3003896899 hasConceptScore W3003896899C154945302 @default.
- W3003896899 hasConceptScore W3003896899C17744445 @default.
- W3003896899 hasConceptScore W3003896899C185798385 @default.
- W3003896899 hasConceptScore W3003896899C199539241 @default.
- W3003896899 hasConceptScore W3003896899C205649164 @default.
- W3003896899 hasConceptScore W3003896899C2776359362 @default.
- W3003896899 hasConceptScore W3003896899C28855332 @default.