Matches in SemOpenAlex for { <https://semopenalex.org/work/W2944859147> ?p ?o ?g. }
- W2944859147 abstract "Existing methods for reducing the computational burden of neural networks at run-time, such as parameter pruning or dynamic computational path selection, focus solely on improving computational efficiency during inference. On the other hand, in this work, we propose a novel method which reduces the memory footprint and number of computing operations required for training and inference. Our framework efficiently integrates pruning as part of the training procedure by exploring and tracking the relative importance of convolutional channels. At each training step, we select only a subset of highly salient channels to execute according to the combinatorial upper confidence bound algorithm, and run a forward and backward pass only on these activated channels, hence learning their parameters. Consequently, we enable the efficient discovery of compact models. We validate our approach empirically on state-of-the-art CNNs - VGGNet, ResNet and DenseNet, and on several image classification datasets. Results demonstrate our framework for dynamic channel execution reduces computational cost up to 4x and parameter count up to 9x, thus reducing the memory and computational demands for discovering and training compact neural network models." @default.
- W2944859147 created "2019-05-29" @default.
- W2944859147 creator A5015529755 @default.
- W2944859147 creator A5056748708 @default.
- W2944859147 date "2019-05-15" @default.
- W2944859147 modified "2023-09-27" @default.
- W2944859147 title "Dynamic Neural Network Channel Execution for Efficient Training" @default.
- W2944859147 cites W1532826273 @default.
- W2944859147 cites W1677182931 @default.
- W2944859147 cites W1686810756 @default.
- W2944859147 cites W1821462560 @default.
- W2944859147 cites W2114766824 @default.
- W2944859147 cites W2167215970 @default.
- W2944859147 cites W2185823609 @default.
- W2944859147 cites W2194775991 @default.
- W2944859147 cites W2260663238 @default.
- W2944859147 cites W2285660444 @default.
- W2944859147 cites W2300242332 @default.
- W2944859147 cites W2335728318 @default.
- W2944859147 cites W2553910756 @default.
- W2944859147 cites W2752037867 @default.
- W2944859147 cites W2783873922 @default.
- W2944859147 cites W2884751099 @default.
- W2944859147 cites W2893585013 @default.
- W2944859147 cites W2899771611 @default.
- W2944859147 cites W2903120399 @default.
- W2944859147 cites W2927152095 @default.
- W2944859147 cites W2962851801 @default.
- W2944859147 cites W2962935523 @default.
- W2944859147 cites W2962944050 @default.
- W2944859147 cites W2962965870 @default.
- W2944859147 cites W2963000224 @default.
- W2944859147 cites W2963048316 @default.
- W2944859147 cites W2963094099 @default.
- W2944859147 cites W2963393494 @default.
- W2944859147 cites W2963446712 @default.
- W2944859147 cites W2963674932 @default.
- W2944859147 cites W2964065888 @default.
- W2944859147 cites W2964118293 @default.
- W2944859147 cites W2964217848 @default.
- W2944859147 cites W3118608800 @default.
- W2944859147 hasPublicationYear "2019" @default.
- W2944859147 type Work @default.
- W2944859147 sameAs 2944859147 @default.
- W2944859147 citedByCount "0" @default.
- W2944859147 crossrefType "posted-content" @default.
- W2944859147 hasAuthorship W2944859147A5015529755 @default.
- W2944859147 hasAuthorship W2944859147A5056748708 @default.
- W2944859147 hasConcept C108010975 @default.
- W2944859147 hasConcept C111335779 @default.
- W2944859147 hasConcept C111919701 @default.
- W2944859147 hasConcept C11413529 @default.
- W2944859147 hasConcept C119857082 @default.
- W2944859147 hasConcept C120665830 @default.
- W2944859147 hasConcept C121332964 @default.
- W2944859147 hasConcept C127162648 @default.
- W2944859147 hasConcept C127964446 @default.
- W2944859147 hasConcept C154945302 @default.
- W2944859147 hasConcept C179799912 @default.
- W2944859147 hasConcept C192209626 @default.
- W2944859147 hasConcept C2524010 @default.
- W2944859147 hasConcept C2776214188 @default.
- W2944859147 hasConcept C31258907 @default.
- W2944859147 hasConcept C33923547 @default.
- W2944859147 hasConcept C41008148 @default.
- W2944859147 hasConcept C50644808 @default.
- W2944859147 hasConcept C6557445 @default.
- W2944859147 hasConcept C74912251 @default.
- W2944859147 hasConcept C81363708 @default.
- W2944859147 hasConcept C86803240 @default.
- W2944859147 hasConceptScore W2944859147C108010975 @default.
- W2944859147 hasConceptScore W2944859147C111335779 @default.
- W2944859147 hasConceptScore W2944859147C111919701 @default.
- W2944859147 hasConceptScore W2944859147C11413529 @default.
- W2944859147 hasConceptScore W2944859147C119857082 @default.
- W2944859147 hasConceptScore W2944859147C120665830 @default.
- W2944859147 hasConceptScore W2944859147C121332964 @default.
- W2944859147 hasConceptScore W2944859147C127162648 @default.
- W2944859147 hasConceptScore W2944859147C127964446 @default.
- W2944859147 hasConceptScore W2944859147C154945302 @default.
- W2944859147 hasConceptScore W2944859147C179799912 @default.
- W2944859147 hasConceptScore W2944859147C192209626 @default.
- W2944859147 hasConceptScore W2944859147C2524010 @default.
- W2944859147 hasConceptScore W2944859147C2776214188 @default.
- W2944859147 hasConceptScore W2944859147C31258907 @default.
- W2944859147 hasConceptScore W2944859147C33923547 @default.
- W2944859147 hasConceptScore W2944859147C41008148 @default.
- W2944859147 hasConceptScore W2944859147C50644808 @default.
- W2944859147 hasConceptScore W2944859147C6557445 @default.
- W2944859147 hasConceptScore W2944859147C74912251 @default.
- W2944859147 hasConceptScore W2944859147C81363708 @default.
- W2944859147 hasConceptScore W2944859147C86803240 @default.
- W2944859147 hasLocation W29448591471 @default.
- W2944859147 hasOpenAccess W2944859147 @default.
- W2944859147 hasPrimaryLocation W29448591471 @default.
- W2944859147 hasRelatedWork W2534788641 @default.
- W2944859147 hasRelatedWork W2807434070 @default.
- W2944859147 hasRelatedWork W2897067384 @default.
- W2944859147 hasRelatedWork W2901523331 @default.
- W2944859147 hasRelatedWork W2905272567 @default.