Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204606360> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W3204606360 endingPage "12238" @default.
- W3204606360 startingPage "12229" @default.
- W3204606360 abstract "Today, state-of-the-art Neural Architecture Search (NAS) methods cannot scale to many hardware platforms or scenarios at a low training costs and/or can only handle non-diverse, heavily constrained architectural search-spaces. To solve these issues, we present DONNA (Distilling Optimal Neural Network Architectures), a novel pipeline for rapid and diverse NAS, that scales to many user scenarios. In DONNA, a search consists of three phases. First, an accuracy predictor is built using blockwise knowledge distillation. This predictor enables searching across diverse networks with varying macro-architectural parameters such as layer types and attention mechanisms as well as across micro-architectural parameters such as block repeats and expansion rates. Second, a rapid evolutionary search phase finds a set of Pareto-optimal architectures for any scenario using the accuracy predictor and on-device measurements. Third, optimal models are quickly finetuned to training-from-scratch accuracy. With this approach, DONNA is up to 100x faster than MNasNet in finding state-of-the-art architectures on-device. Classifying ImageNet, DONNA architectures are 20% faster than EfficientNet-B0 and MobileNetV2 on a Nvidia V100 GPU and 10% faster with 0.5% higher accuracy than MobileNetV2-1.4x on a Samsung S20 smartphone. In addition to NAS, DONNA is used for search-space extension and exploration, as well as hardware-aware model compression." @default.
- W3204606360 created "2021-10-11" @default.
- W3204606360 creator A5013373065 @default.
- W3204606360 creator A5024419624 @default.
- W3204606360 creator A5028443888 @default.
- W3204606360 creator A5040865542 @default.
- W3204606360 creator A5061910461 @default.
- W3204606360 creator A5075709004 @default.
- W3204606360 creator A5082671220 @default.
- W3204606360 date "2020-01-01" @default.
- W3204606360 modified "2023-09-27" @default.
- W3204606360 title "Distilling Optimal Neural Networks: Rapid Search in Diverse Spaces." @default.
- W3204606360 hasPublicationYear "2020" @default.
- W3204606360 type Work @default.
- W3204606360 sameAs 3204606360 @default.
- W3204606360 citedByCount "0" @default.
- W3204606360 crossrefType "proceedings-article" @default.
- W3204606360 hasAuthorship W3204606360A5013373065 @default.
- W3204606360 hasAuthorship W3204606360A5024419624 @default.
- W3204606360 hasAuthorship W3204606360A5028443888 @default.
- W3204606360 hasAuthorship W3204606360A5040865542 @default.
- W3204606360 hasAuthorship W3204606360A5061910461 @default.
- W3204606360 hasAuthorship W3204606360A5075709004 @default.
- W3204606360 hasAuthorship W3204606360A5082671220 @default.
- W3204606360 hasConcept C113775141 @default.
- W3204606360 hasConcept C118524514 @default.
- W3204606360 hasConcept C119857082 @default.
- W3204606360 hasConcept C123657996 @default.
- W3204606360 hasConcept C142362112 @default.
- W3204606360 hasConcept C153349607 @default.
- W3204606360 hasConcept C154945302 @default.
- W3204606360 hasConcept C177264268 @default.
- W3204606360 hasConcept C199360897 @default.
- W3204606360 hasConcept C2524010 @default.
- W3204606360 hasConcept C2777210771 @default.
- W3204606360 hasConcept C33923547 @default.
- W3204606360 hasConcept C41008148 @default.
- W3204606360 hasConcept C43521106 @default.
- W3204606360 hasConcept C50644808 @default.
- W3204606360 hasConceptScore W3204606360C113775141 @default.
- W3204606360 hasConceptScore W3204606360C118524514 @default.
- W3204606360 hasConceptScore W3204606360C119857082 @default.
- W3204606360 hasConceptScore W3204606360C123657996 @default.
- W3204606360 hasConceptScore W3204606360C142362112 @default.
- W3204606360 hasConceptScore W3204606360C153349607 @default.
- W3204606360 hasConceptScore W3204606360C154945302 @default.
- W3204606360 hasConceptScore W3204606360C177264268 @default.
- W3204606360 hasConceptScore W3204606360C199360897 @default.
- W3204606360 hasConceptScore W3204606360C2524010 @default.
- W3204606360 hasConceptScore W3204606360C2777210771 @default.
- W3204606360 hasConceptScore W3204606360C33923547 @default.
- W3204606360 hasConceptScore W3204606360C41008148 @default.
- W3204606360 hasConceptScore W3204606360C43521106 @default.
- W3204606360 hasConceptScore W3204606360C50644808 @default.
- W3204606360 hasLocation W32046063601 @default.
- W3204606360 hasOpenAccess W3204606360 @default.
- W3204606360 hasPrimaryLocation W32046063601 @default.
- W3204606360 hasRelatedWork W2748513770 @default.
- W3204606360 hasRelatedWork W2901177033 @default.
- W3204606360 hasRelatedWork W2957020430 @default.
- W3204606360 hasRelatedWork W3008968734 @default.
- W3204606360 hasRelatedWork W3014859179 @default.
- W3204606360 hasRelatedWork W3090402053 @default.
- W3204606360 hasRelatedWork W3091716143 @default.
- W3204606360 hasRelatedWork W3094590205 @default.
- W3204606360 hasRelatedWork W3104482253 @default.
- W3204606360 hasRelatedWork W3109915984 @default.
- W3204606360 hasRelatedWork W3110460514 @default.
- W3204606360 hasRelatedWork W3110978130 @default.
- W3204606360 hasRelatedWork W3111941549 @default.
- W3204606360 hasRelatedWork W3175549063 @default.
- W3204606360 hasRelatedWork W3184682079 @default.
- W3204606360 hasRelatedWork W3187682281 @default.
- W3204606360 hasRelatedWork W3200067639 @default.
- W3204606360 hasRelatedWork W3200647350 @default.
- W3204606360 hasRelatedWork W3209022794 @default.
- W3204606360 hasRelatedWork W2971428146 @default.
- W3204606360 isParatext "false" @default.
- W3204606360 isRetracted "false" @default.
- W3204606360 magId "3204606360" @default.
- W3204606360 workType "article" @default.