FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2513302235> ?p ?o ?g. }

• W2513302235 abstract "The computation and storage requirements for Deep Neural Networks (DNNs) are usually high. This issue limits their deployability on ubiquitous computing devices such as smart phones, wearables and autonomous drones. In this paper, we propose ternary neural networks (TNNs) in order to make deep learning more resource-efficient. We train these TNNs using a teacher-student approach based on a novel, layer-wise greedy methodology. Thanks to our two-stage training procedure, the teacher network is still able to use state-of-the-art methods such as dropout and batch normalization to increase accuracy and reduce training time. Using only ternary weights and activations, the student ternary network learns to mimic the behavior of its teacher network without using any multiplication. Unlike its -1,1 binary counterparts, a ternary neural network inherently prunes the smaller weights by setting them to zero during training. This makes them sparser and thus more energy-efficient. We design a purpose-built hardware architecture for TNNs and implement it on FPGA and ASIC. We evaluate TNNs on several benchmark datasets and demonstrate up to 3.1x better energy efficiency with respect to the state of the art while also improving accuracy." @default.
• W2513302235 created "2016-09-16" @default.
• W2513302235 creator A5049495734 @default.
• W2513302235 creator A5068370976 @default.
• W2513302235 creator A5085830369 @default.
• W2513302235 creator A5088674104 @default.
• W2513302235 date "2016-09-01" @default.
• W2513302235 modified "2023-09-30" @default.
• W2513302235 title "Ternary Neural Networks for Resource-Efficient AI Applications" @default.
• W2513302235 cites W1649885719 @default.
• W2513302235 cites W1701967867 @default.
• W2513302235 cites W1902934009 @default.
• W2513302235 cites W2067713319 @default.
• W2513302235 cites W2095705004 @default.
• W2513302235 cites W2112796928 @default.
• W2513302235 cites W2138913040 @default.
• W2513302235 cites W2143612262 @default.
• W2513302235 cites W2161758346 @default.
• W2513302235 cites W2266701264 @default.
• W2513302235 cites W2314470091 @default.
• W2513302235 cites W2319920447 @default.
• W2513302235 cites W2335728318 @default.
• W2513302235 cites W2405102949 @default.
• W2513302235 cites W2405920868 @default.
• W2513302235 cites W2469490737 @default.
• W2513302235 cites W2949117887 @default.
• W2513302235 cites W2950894517 @default.
• W2513302235 cites W2964200805 @default.
• W2513302235 cites W2964299589 @default.
• W2513302235 cites W2964308564 @default.
• W2513302235 cites W3118608800 @default.
• W2513302235 hasPublicationYear "2016" @default.
• W2513302235 type Work @default.
• W2513302235 sameAs 2513302235 @default.
• W2513302235 citedByCount "6" @default.
• W2513302235 countsByYear W25133022352017 @default.
• W2513302235 countsByYear W25133022352018 @default.
• W2513302235 countsByYear W25133022352019 @default.
• W2513302235 countsByYear W25133022352021 @default.
• W2513302235 crossrefType "posted-content" @default.
• W2513302235 hasAuthorship W2513302235A5049495734 @default.
• W2513302235 hasAuthorship W2513302235A5068370976 @default.
• W2513302235 hasAuthorship W2513302235A5085830369 @default.
• W2513302235 hasAuthorship W2513302235A5088674104 @default.
• W2513302235 hasConcept C113775141 @default.
• W2513302235 hasConcept C11413529 @default.
• W2513302235 hasConcept C119857082 @default.
• W2513302235 hasConcept C13280743 @default.
• W2513302235 hasConcept C136886441 @default.
• W2513302235 hasConcept C144024400 @default.
• W2513302235 hasConcept C149635348 @default.
• W2513302235 hasConcept C154945302 @default.
• W2513302235 hasConcept C185798385 @default.
• W2513302235 hasConcept C19165224 @default.
• W2513302235 hasConcept C199360897 @default.
• W2513302235 hasConcept C205649164 @default.
• W2513302235 hasConcept C2776145597 @default.
• W2513302235 hasConcept C41008148 @default.
• W2513302235 hasConcept C42935608 @default.
• W2513302235 hasConcept C45374587 @default.
• W2513302235 hasConcept C50644808 @default.
• W2513302235 hasConcept C64452783 @default.
• W2513302235 hasConcept C77390884 @default.
• W2513302235 hasConceptScore W2513302235C113775141 @default.
• W2513302235 hasConceptScore W2513302235C11413529 @default.
• W2513302235 hasConceptScore W2513302235C119857082 @default.
• W2513302235 hasConceptScore W2513302235C13280743 @default.
• W2513302235 hasConceptScore W2513302235C136886441 @default.
• W2513302235 hasConceptScore W2513302235C144024400 @default.
• W2513302235 hasConceptScore W2513302235C149635348 @default.
• W2513302235 hasConceptScore W2513302235C154945302 @default.
• W2513302235 hasConceptScore W2513302235C185798385 @default.
• W2513302235 hasConceptScore W2513302235C19165224 @default.
• W2513302235 hasConceptScore W2513302235C199360897 @default.
• W2513302235 hasConceptScore W2513302235C205649164 @default.
• W2513302235 hasConceptScore W2513302235C2776145597 @default.
• W2513302235 hasConceptScore W2513302235C41008148 @default.
• W2513302235 hasConceptScore W2513302235C42935608 @default.
• W2513302235 hasConceptScore W2513302235C45374587 @default.
• W2513302235 hasConceptScore W2513302235C50644808 @default.
• W2513302235 hasConceptScore W2513302235C64452783 @default.
• W2513302235 hasConceptScore W2513302235C77390884 @default.
• W2513302235 hasOpenAccess W2513302235 @default.
• W2513302235 hasRelatedWork W2048266589 @default.
• W2513302235 hasRelatedWork W2585560244 @default.
• W2513302235 hasRelatedWork W2606722458 @default.
• W2513302235 hasRelatedWork W2775600882 @default.
• W2513302235 hasRelatedWork W2897411032 @default.
• W2513302235 hasRelatedWork W2897655610 @default.
• W2513302235 hasRelatedWork W2903990228 @default.
• W2513302235 hasRelatedWork W2918653705 @default.
• W2513302235 hasRelatedWork W2951876100 @default.
• W2513302235 hasRelatedWork W2963536136 @default.
• W2513302235 hasRelatedWork W2964299589 @default.
• W2513302235 hasRelatedWork W2969396878 @default.
• W2513302235 hasRelatedWork W2977748703 @default.
• W2513302235 hasRelatedWork W2981240861 @default.
• W2513302235 hasRelatedWork W3116954976 @default.
• W2513302235 hasRelatedWork W3123482262 @default.
• W2513302235 hasRelatedWork W3126583178 @default.

• next