FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3199014737> ?p ?o ?g. }

• W3199014737 abstract "Quantizing deep networks with adaptive bit-widths is a promising technique for efficient inference across many devices and resource constraints. In contrast to static methods that repeat the quantization process and train different models for different constraints, adaptive quantization enables us to flexibly adjust the bit-widths of a single deep network during inference for instant adaptation in different scenarios. While existing research shows encouraging results on common image classification benchmarks, this paper investigates how to train such adaptive networks more effectively. Specifically, we present two novel techniques for quantizing deep neural networks with adaptive bit-widths of weights and activations. First, we propose a collaborative strategy to choose a high-precision teacher for transferring knowledge to the low-precision student while jointly optimizing the model with all bit-widths. Second, to effectively transfer knowledge, we develop a dynamic block swapping method by randomly replacing the blocks in the lower-precision student network with the corresponding blocks in the higher-precision teacher network. Extensive experiments on multiple image classification datasets including video classification benchmarks for the first time, well demonstrate the efficacy of our approach over state-of-the-art methods." @default.
• W3199014737 created "2021-09-27" @default.
• W3199014737 creator A5004766079 @default.
• W3199014737 creator A5016934680 @default.
• W3199014737 creator A5021257546 @default.
• W3199014737 creator A5049734237 @default.
• W3199014737 creator A5052325109 @default.
• W3199014737 creator A5075906727 @default.
• W3199014737 creator A5078843154 @default.
• W3199014737 creator A5080417608 @default.
• W3199014737 creator A5082043392 @default.
• W3199014737 date "2021-03-02" @default.
• W3199014737 modified "2023-09-27" @default.
• W3199014737 title "Improved Techniques for Quantizing Deep Networks with Adaptive Bit-Widths" @default.
• W3199014737 cites W1821462560 @default.
• W3199014737 cites W1927052826 @default.
• W3199014737 cites W2179423374 @default.
• W3199014737 cites W2242818861 @default.
• W3199014737 cites W2267635276 @default.
• W3199014737 cites W2296073425 @default.
• W3199014737 cites W2300242332 @default.
• W3199014737 cites W2405920868 @default.
• W3199014737 cites W2469490737 @default.
• W3199014737 cites W2507009361 @default.
• W3199014737 cites W2586654419 @default.
• W3199014737 cites W2613718673 @default.
• W3199014737 cites W2620998106 @default.
• W3199014737 cites W2767421475 @default.
• W3199014737 cites W2786771851 @default.
• W3199014737 cites W2787017828 @default.
• W3199014737 cites W2796394093 @default.
• W3199014737 cites W2884150179 @default.
• W3199014737 cites W2903260438 @default.
• W3199014737 cites W2924561803 @default.
• W3199014737 cites W2927589347 @default.
• W3199014737 cites W2949829435 @default.
• W3199014737 cites W2962944050 @default.
• W3199014737 cites W2963140444 @default.
• W3199014737 cites W2963163009 @default.
• W3199014737 cites W2963373431 @default.
• W3199014737 cites W2963524571 @default.
• W3199014737 cites W2963877604 @default.
• W3199014737 cites W2963897760 @default.
• W3199014737 cites W2964118293 @default.
• W3199014737 cites W2964220233 @default.
• W3199014737 cites W2970803838 @default.
• W3199014737 cites W2982479999 @default.
• W3199014737 cites W2990953743 @default.
• W3199014737 cites W2994749257 @default.
• W3199014737 cites W2997006708 @default.
• W3199014737 cites W3004127093 @default.
• W3199014737 cites W3034644181 @default.
• W3199014737 cites W3034887213 @default.
• W3199014737 cites W3035321581 @default.
• W3199014737 cites W3040128780 @default.
• W3199014737 cites W3101155878 @default.
• W3199014737 cites W3108411658 @default.
• W3199014737 cites W3108833523 @default.
• W3199014737 cites W3118608800 @default.
• W3199014737 cites W2426267443 @default.
• W3199014737 hasPublicationYear "2021" @default.
• W3199014737 type Work @default.
• W3199014737 sameAs 3199014737 @default.
• W3199014737 citedByCount "0" @default.
• W3199014737 crossrefType "posted-content" @default.
• W3199014737 hasAuthorship W3199014737A5004766079 @default.
• W3199014737 hasAuthorship W3199014737A5016934680 @default.
• W3199014737 hasAuthorship W3199014737A5021257546 @default.
• W3199014737 hasAuthorship W3199014737A5049734237 @default.
• W3199014737 hasAuthorship W3199014737A5052325109 @default.
• W3199014737 hasAuthorship W3199014737A5075906727 @default.
• W3199014737 hasAuthorship W3199014737A5078843154 @default.
• W3199014737 hasAuthorship W3199014737A5080417608 @default.
• W3199014737 hasAuthorship W3199014737A5082043392 @default.
• W3199014737 hasConcept C113775141 @default.
• W3199014737 hasConcept C11413529 @default.
• W3199014737 hasConcept C119857082 @default.
• W3199014737 hasConcept C154945302 @default.
• W3199014737 hasConcept C2776214188 @default.
• W3199014737 hasConcept C28855332 @default.
• W3199014737 hasConcept C2984842247 @default.
• W3199014737 hasConcept C41008148 @default.
• W3199014737 hasConcept C50644808 @default.
• W3199014737 hasConceptScore W3199014737C113775141 @default.
• W3199014737 hasConceptScore W3199014737C11413529 @default.
• W3199014737 hasConceptScore W3199014737C119857082 @default.
• W3199014737 hasConceptScore W3199014737C154945302 @default.
• W3199014737 hasConceptScore W3199014737C2776214188 @default.
• W3199014737 hasConceptScore W3199014737C28855332 @default.
• W3199014737 hasConceptScore W3199014737C2984842247 @default.
• W3199014737 hasConceptScore W3199014737C41008148 @default.
• W3199014737 hasConceptScore W3199014737C50644808 @default.
• W3199014737 hasLocation W31990147371 @default.
• W3199014737 hasOpenAccess W3199014737 @default.
• W3199014737 hasPrimaryLocation W31990147371 @default.
• W3199014737 hasRelatedWork W2786771851 @default.
• W3199014737 hasRelatedWork W2902355249 @default.
• W3199014737 hasRelatedWork W2904640605 @default.
• W3199014737 hasRelatedWork W2950297794 @default.
• W3199014737 hasRelatedWork W2962298324 @default.

• next