FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2912648642> ?p ?o ?g. }

• W2912648642 abstract "The success of convolutional neural networks (CNNs) in computer vision applications has been accompanied by a significant increase of computation and memory costs, which prohibits its usage on resource-limited environments such as mobile or embedded devices. To this end, the research of CNN compression has recently become emerging. In this paper, we propose a novel filter pruning scheme, termed structured sparsity regularization (SSR), to simultaneously speedup the computation and reduce the memory overhead of CNNs, which can be well supported by various off-the-shelf deep learning libraries. Concretely, the proposed scheme incorporates two different regularizers of structured sparsity into the original objective function of filter pruning, which fully coordinates the global outputs and local pruning operations to adaptively prune filters. We further propose an Alternative Updating with Lagrange Multipliers (AULM) scheme to efficiently solve its optimization. AULM follows the principle of ADMM and alternates between promoting the structured sparsity of CNNs and optimizing the recognition loss, which leads to a very efficient solver (2.5x to the most recent work that directly solves the group sparsity-based regularization). Moreover, by imposing the structured sparsity, the online inference is extremely memory-light, since the number of filters and the output feature maps are simultaneously reduced. The proposed scheme has been deployed to a variety of state-of-the-art CNN structures including LeNet, AlexNet, VGG, ResNet and GoogLeNet over different datasets. Quantitative results demonstrate that the proposed scheme achieves superior performance over the state-of-the-art methods. We further demonstrate the proposed compression scheme for the task of transfer learning, including domain adaptation and object detection, which also show exciting performance gains over the state-of-the-arts." @default.
• W2912648642 created "2019-02-21" @default.
• W2912648642 creator A5015874725 @default.
• W2912648642 creator A5016080094 @default.
• W2912648642 creator A5043643513 @default.
• W2912648642 creator A5068918243 @default.
• W2912648642 creator A5070551517 @default.
• W2912648642 date "2019-01-23" @default.
• W2912648642 modified "2023-09-26" @default.
• W2912648642 title "Towards Compact ConvNets via Structure-Sparsity Regularized Filter Pruning" @default.
• W2912648642 cites W1536680647 @default.
• W2912648642 cites W1665214252 @default.
• W2912648642 cites W1686810756 @default.
• W2912648642 cites W1724438581 @default.
• W2912648642 cites W1789336918 @default.
• W2912648642 cites W1871489475 @default.
• W2912648642 cites W1922123711 @default.
• W2912648642 cites W1935978687 @default.
• W2912648642 cites W2097117768 @default.
• W2912648642 cites W2102605133 @default.
• W2912648642 cites W2112796928 @default.
• W2912648642 cites W2117539524 @default.
• W2912648642 cites W2125389748 @default.
• W2912648642 cites W2131524184 @default.
• W2912648642 cites W2138019504 @default.
• W2912648642 cites W2155893237 @default.
• W2912648642 cites W2161591461 @default.
• W2912648642 cites W2162409952 @default.
• W2912648642 cites W2163605009 @default.
• W2912648642 cites W2163922914 @default.
• W2912648642 cites W2164278908 @default.
• W2912648642 cites W2167215970 @default.
• W2912648642 cites W2194775991 @default.
• W2912648642 cites W2233116163 @default.
• W2912648642 cites W2260663238 @default.
• W2912648642 cites W2271840356 @default.
• W2912648642 cites W2279098554 @default.
• W2912648642 cites W2285660444 @default.
• W2912648642 cites W2300242332 @default.
• W2912648642 cites W2319920447 @default.
• W2912648642 cites W2495425901 @default.
• W2912648642 cites W2531409750 @default.
• W2912648642 cites W2549139847 @default.
• W2912648642 cites W2554242204 @default.
• W2912648642 cites W2554302513 @default.
• W2912648642 cites W2575345876 @default.
• W2912648642 cites W2612445135 @default.
• W2912648642 cites W2613718673 @default.
• W2912648642 cites W2741485222 @default.
• W2912648642 cites W2758000438 @default.
• W2912648642 cites W2764289073 @default.
• W2912648642 cites W2775811337 @default.
• W2912648642 cites W2789135445 @default.
• W2912648642 cites W2790852735 @default.
• W2912648642 cites W2807961551 @default.
• W2912648642 cites W2895561155 @default.
• W2912648642 cites W2913081068 @default.
• W2912648642 cites W2949117887 @default.
• W2912648642 cites W2950967261 @default.
• W2912648642 cites W2951977814 @default.
• W2912648642 cites W2952826672 @default.
• W2912648642 cites W2962851801 @default.
• W2912648642 cites W2962965870 @default.
• W2912648642 cites W2962988160 @default.
• W2912648642 cites W2963000224 @default.
• W2912648642 cites W2963114950 @default.
• W2912648642 cites W2963125010 @default.
• W2912648642 cites W2963163009 @default.
• W2912648642 cites W2963225922 @default.
• W2912648642 cites W2963287528 @default.
• W2912648642 cites W2963446712 @default.
• W2912648642 cites W2963674932 @default.
• W2912648642 cites W2963911037 @default.
• W2912648642 cites W2963993763 @default.
• W2912648642 cites W2964233199 @default.
• W2912648642 cites W2964299589 @default.
• W2912648642 cites W3106250896 @default.
• W2912648642 cites W566555209 @default.
• W2912648642 cites W992687842 @default.
• W2912648642 cites W14333344 @default.
• W2912648642 hasPublicationYear "2019" @default.
• W2912648642 type Work @default.
• W2912648642 sameAs 2912648642 @default.
• W2912648642 citedByCount "5" @default.
• W2912648642 countsByYear W29126486422019 @default.
• W2912648642 countsByYear W29126486422020 @default.
• W2912648642 countsByYear W29126486422021 @default.
• W2912648642 crossrefType "posted-content" @default.
• W2912648642 hasAuthorship W2912648642A5015874725 @default.
• W2912648642 hasAuthorship W2912648642A5016080094 @default.
• W2912648642 hasAuthorship W2912648642A5043643513 @default.
• W2912648642 hasAuthorship W2912648642A5068918243 @default.
• W2912648642 hasAuthorship W2912648642A5070551517 @default.
• W2912648642 hasConcept C106131492 @default.
• W2912648642 hasConcept C108010975 @default.
• W2912648642 hasConcept C111919701 @default.
• W2912648642 hasConcept C11413529 @default.
• W2912648642 hasConcept C138885662 @default.
• W2912648642 hasConcept C153180895 @default.
• W2912648642 hasConcept C154945302 @default.

• next