FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3174892017> ?p ?o ?g. }

• W3174892017 abstract "In this paper, we propose a novel layer based on fast Walsh-Hadamard transform (WHT) and smooth-thresholding to replace 1 × 1 convolution layers in deep neural networks. In the WHT domain, we denoise the transform domain coefficients using the new smooth-thresholding non-linearity, a smoothed version of the well-known soft-thresholding operator. We also introduce a family of multiplication-free operators from the basic 2×2 Hadamard transform to implement 3 × 3 depthwise separable convolution layers. Using these two types of layers, we replace the bottleneck layers in MobileNet-V2 to reduce the network’s number of parameters with a slight loss in accuracy. For example, by replacing the final third bottleneck layers, we reduce the number of parameters from 2.270M to 947K. This reduces the accuracy from 95.21% to 92.88% on the CIFAR-10 dataset. Our approach significantly improves the speed of data processing. The fast Walsh-Hadamard transform has a computational complexity of O(m log <inf xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>2</inf> m). As a result, it is computationally more efficient than the 1 × 1 convolution layer. The fast Walsh-Hadamard layer processes a tensor in ℝ <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>10×32×32×1024</sup> about 2 times faster than 1 × 1 convolution layer on NVIDIA Jetson Nano computer board." @default.
• W3174892017 created "2021-07-05" @default.
• W3174892017 creator A5002608156 @default.
• W3174892017 creator A5058826860 @default.
• W3174892017 creator A5080469744 @default.
• W3174892017 date "2021-06-01" @default.
• W3174892017 modified "2023-09-29" @default.
• W3174892017 title "Fast Walsh-Hadamard Transform and Smooth-Thresholding Based Binary Layers in Deep Neural Networks" @default.
• W3174892017 cites W1779006005 @default.
• W3174892017 cites W1903029394 @default.
• W3174892017 cites W1925312670 @default.
• W3174892017 cites W2097117768 @default.
• W3174892017 cites W2146842127 @default.
• W3174892017 cites W2164921999 @default.
• W3174892017 cites W2168587114 @default.
• W3174892017 cites W2194775991 @default.
• W3174892017 cites W2233116163 @default.
• W3174892017 cites W2322432035 @default.
• W3174892017 cites W2333786694 @default.
• W3174892017 cites W2549139847 @default.
• W3174892017 cites W2611702063 @default.
• W3174892017 cites W2732618413 @default.
• W3174892017 cites W2885527679 @default.
• W3174892017 cites W2885647100 @default.
• W3174892017 cites W2892104683 @default.
• W3174892017 cites W2940390282 @default.
• W3174892017 cites W2963163009 @default.
• W3174892017 cites W2963495494 @default.
• W3174892017 cites W2964350391 @default.
• W3174892017 cites W2972442919 @default.
• W3174892017 cites W2973854309 @default.
• W3174892017 cites W2981689412 @default.
• W3174892017 cites W2985488694 @default.
• W3174892017 cites W3003910221 @default.
• W3174892017 cites W3007570526 @default.
• W3174892017 cites W3011862431 @default.
• W3174892017 cites W3015879607 @default.
• W3174892017 cites W3015887714 @default.
• W3174892017 cites W3026536359 @default.
• W3174892017 cites W3106941003 @default.
• W3174892017 cites W3108720720 @default.
• W3174892017 cites W3133262691 @default.
• W3174892017 cites W3170336723 @default.
• W3174892017 doi "https://doi.org/10.1109/cvprw53098.2021.00523" @default.
• W3174892017 hasPublicationYear "2021" @default.
• W3174892017 type Work @default.
• W3174892017 sameAs 3174892017 @default.
• W3174892017 citedByCount "7" @default.
• W3174892017 countsByYear W31748920172022 @default.
• W3174892017 countsByYear W31748920172023 @default.
• W3174892017 crossrefType "proceedings-article" @default.
• W3174892017 hasAuthorship W3174892017A5002608156 @default.
• W3174892017 hasAuthorship W3174892017A5058826860 @default.
• W3174892017 hasAuthorship W3174892017A5080469744 @default.
• W3174892017 hasBestOaLocation W31748920172 @default.
• W3174892017 hasConcept C11413529 @default.
• W3174892017 hasConcept C115961682 @default.
• W3174892017 hasConcept C134306372 @default.
• W3174892017 hasConcept C149635348 @default.
• W3174892017 hasConcept C154945302 @default.
• W3174892017 hasConcept C164319782 @default.
• W3174892017 hasConcept C191178318 @default.
• W3174892017 hasConcept C2780513914 @default.
• W3174892017 hasConcept C33923547 @default.
• W3174892017 hasConcept C41008148 @default.
• W3174892017 hasConcept C45347329 @default.
• W3174892017 hasConcept C50644808 @default.
• W3174892017 hasConcept C60292330 @default.
• W3174892017 hasConceptScore W3174892017C11413529 @default.
• W3174892017 hasConceptScore W3174892017C115961682 @default.
• W3174892017 hasConceptScore W3174892017C134306372 @default.
• W3174892017 hasConceptScore W3174892017C149635348 @default.
• W3174892017 hasConceptScore W3174892017C154945302 @default.
• W3174892017 hasConceptScore W3174892017C164319782 @default.
• W3174892017 hasConceptScore W3174892017C191178318 @default.
• W3174892017 hasConceptScore W3174892017C2780513914 @default.
• W3174892017 hasConceptScore W3174892017C33923547 @default.
• W3174892017 hasConceptScore W3174892017C41008148 @default.
• W3174892017 hasConceptScore W3174892017C45347329 @default.
• W3174892017 hasConceptScore W3174892017C50644808 @default.
• W3174892017 hasConceptScore W3174892017C60292330 @default.
• W3174892017 hasLocation W31748920171 @default.
• W3174892017 hasLocation W31748920172 @default.
• W3174892017 hasOpenAccess W3174892017 @default.
• W3174892017 hasPrimaryLocation W31748920171 @default.
• W3174892017 hasRelatedWork W1846298089 @default.
• W3174892017 hasRelatedWork W2034089923 @default.
• W3174892017 hasRelatedWork W2035171694 @default.
• W3174892017 hasRelatedWork W2061222953 @default.
• W3174892017 hasRelatedWork W2104099160 @default.
• W3174892017 hasRelatedWork W2143999670 @default.
• W3174892017 hasRelatedWork W2363656651 @default.
• W3174892017 hasRelatedWork W3174892017 @default.
• W3174892017 hasRelatedWork W4225282419 @default.
• W3174892017 hasRelatedWork W2020400385 @default.
• W3174892017 isParatext "false" @default.
• W3174892017 isRetracted "false" @default.
• W3174892017 magId "3174892017" @default.
• W3174892017 workType "article" @default.