FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2892786111> ?p ?o ?g. }

• W2892786111 abstract "We propose to execute deep neural networks (DNNs) with dynamic and sparse graph (DSG) structure for compressive memory and accelerative execution during both training and inference. The great success of DNNs motivates the pursuing of lightweight models for the deployment onto embedded devices. However, most of the previous studies optimize for inference while neglect training or even complicate it. Training is far more intractable, since (i) the neurons dominate the memory cost rather than the weights in inference; (ii) the dynamic activation makes previous sparse acceleration via one-off optimization on fixed weight invalid; (iii) batch normalization (BN) is critical for maintaining accuracy while its activation reorganization damages the sparsity. To address these issues, DSG activates only a small amount of neurons with high selectivity at each iteration via a dimension-reduction search (DRS) and obtains the BN compatibility via a double-mask selection (DMS). Experiments show significant memory saving (1.7-4.5x) and operation reduction (2.3-4.4x) with little accuracy loss on various benchmarks." @default.
• W2892786111 created "2018-10-05" @default.
• W2892786111 creator A5004075655 @default.
• W2892786111 creator A5018970859 @default.
• W2892786111 creator A5047545398 @default.
• W2892786111 creator A5048052285 @default.
• W2892786111 creator A5063355700 @default.
• W2892786111 creator A5082076121 @default.
• W2892786111 creator A5082604341 @default.
• W2892786111 date "2018-10-01" @default.
• W2892786111 modified "2023-09-27" @default.
• W2892786111 title "Dynamic Sparse Graph for Efficient Deep Learning" @default.
• W2892786111 cites W1686810756 @default.
• W2892786111 cites W1798945469 @default.
• W2892786111 cites W1995562189 @default.
• W2892786111 cites W2041836310 @default.
• W2892786111 cites W2053171205 @default.
• W2892786111 cites W2056738732 @default.
• W2892786111 cites W2089497633 @default.
• W2892786111 cites W2093813380 @default.
• W2892786111 cites W2108598243 @default.
• W2892786111 cites W2112796928 @default.
• W2892786111 cites W2132153846 @default.
• W2892786111 cites W2142119745 @default.
• W2892786111 cites W2178237821 @default.
• W2892786111 cites W2194775991 @default.
• W2892786111 cites W2279057335 @default.
• W2892786111 cites W2285660444 @default.
• W2892786111 cites W2319920447 @default.
• W2892786111 cites W2401231614 @default.
• W2892786111 cites W2469490737 @default.
• W2892786111 cites W2515385951 @default.
• W2892786111 cites W2525778437 @default.
• W2892786111 cites W2550815077 @default.
• W2892786111 cites W2559813832 @default.
• W2892786111 cites W2570343428 @default.
• W2892786111 cites W2585720638 @default.
• W2892786111 cites W2593052414 @default.
• W2892786111 cites W2622263826 @default.
• W2892786111 cites W2707890836 @default.
• W2892786111 cites W2733236492 @default.
• W2892786111 cites W2739789140 @default.
• W2892786111 cites W2750384547 @default.
• W2892786111 cites W2755682530 @default.
• W2892786111 cites W2766164908 @default.
• W2892786111 cites W2785648721 @default.
• W2892786111 cites W2786054724 @default.
• W2892786111 cites W2805933112 @default.
• W2892786111 cites W2883283076 @default.
• W2892786111 cites W2884180697 @default.
• W2892786111 cites W2886851211 @default.
• W2892786111 cites W2890616119 @default.
• W2892786111 cites W2893585013 @default.
• W2892786111 cites W2919115771 @default.
• W2892786111 cites W2949117887 @default.
• W2892786111 cites W2951977814 @default.
• W2892786111 cites W2962821792 @default.
• W2892786111 cites W2962851801 @default.
• W2892786111 cites W2962939807 @default.
• W2892786111 cites W2963000224 @default.
• W2892786111 cites W2963363373 @default.
• W2892786111 cites W2963674932 @default.
• W2892786111 cites W2963803379 @default.
• W2892786111 cites W2964004663 @default.
• W2892786111 cites W2964152344 @default.
• W2892786111 cites W2964164125 @default.
• W2892786111 cites W2964233199 @default.
• W2892786111 cites W2964299589 @default.
• W2892786111 cites W2979473749 @default.
• W2892786111 cites W3028304412 @default.
• W2892786111 cites W3118608800 @default.
• W2892786111 cites W2701971652 @default.
• W2892786111 hasPublicationYear "2018" @default.
• W2892786111 type Work @default.
• W2892786111 sameAs 2892786111 @default.
• W2892786111 citedByCount "3" @default.
• W2892786111 countsByYear W28927861112019 @default.
• W2892786111 countsByYear W28927861112020 @default.
• W2892786111 countsByYear W28927861112021 @default.
• W2892786111 crossrefType "posted-content" @default.
• W2892786111 hasAuthorship W2892786111A5004075655 @default.
• W2892786111 hasAuthorship W2892786111A5018970859 @default.
• W2892786111 hasAuthorship W2892786111A5047545398 @default.
• W2892786111 hasAuthorship W2892786111A5048052285 @default.
• W2892786111 hasAuthorship W2892786111A5063355700 @default.
• W2892786111 hasAuthorship W2892786111A5082076121 @default.
• W2892786111 hasAuthorship W2892786111A5082604341 @default.
• W2892786111 hasConcept C119857082 @default.
• W2892786111 hasConcept C132525143 @default.
• W2892786111 hasConcept C136886441 @default.
• W2892786111 hasConcept C144024400 @default.
• W2892786111 hasConcept C154945302 @default.
• W2892786111 hasConcept C19165224 @default.
• W2892786111 hasConcept C2776214188 @default.
• W2892786111 hasConcept C2984842247 @default.
• W2892786111 hasConcept C41008148 @default.
• W2892786111 hasConcept C50644808 @default.
• W2892786111 hasConcept C80444323 @default.
• W2892786111 hasConceptScore W2892786111C119857082 @default.
• W2892786111 hasConceptScore W2892786111C132525143 @default.

• next