FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2946506871> ?p ?o ?g. }

• W2946506871 abstract "Convolutional neural networks (CNNs) achieve state-of-the-art accuracy in a variety of tasks in computer vision and beyond. One of the major obstacles hindering the ubiquitous use of CNNs for inference on low-power edge devices is their high computational complexity and memory bandwidth requirements. The latter often dominates the energy footprint on modern hardware. In this paper, we introduce a lossy transform coding approach, inspired by image and video compression, designed to reduce the memory bandwidth due to the storage of intermediate activation calculation results. Our method does not require fine-tuning the network weights and halves the data transfer volumes to the main memory by compressing feature maps, which are highly correlated, with variable length coding. Our method outperform previous approach in term of the number of bits per value with minor accuracy degradation on ResNet-34 and MobileNetV2. We analyze the performance of our approach on a variety of CNN architectures and demonstrate that FPGA implementation of ResNet-18 with our approach results in a reduction of around 40% in the memory energy footprint, compared to quantized network, with negligible impact on accuracy. When allowing accuracy degradation of up to 2%, the reduction of 60% is achieved. A reference implementation is available at this https URL" @default.
• W2946506871 created "2019-05-29" @default.
• W2946506871 creator A5017964819 @default.
• W2946506871 creator A5019913171 @default.
• W2946506871 creator A5023269428 @default.
• W2946506871 creator A5025410418 @default.
• W2946506871 creator A5029515956 @default.
• W2946506871 creator A5031600563 @default.
• W2946506871 creator A5089135250 @default.
• W2946506871 creator A5091675091 @default.
• W2946506871 date "2019-05-26" @default.
• W2946506871 modified "2023-09-27" @default.
• W2946506871 title "Feature Map Transform Coding for Energy-Efficient CNN Inference." @default.
• W2946506871 cites W2097117768 @default.
• W2946506871 cites W2108277362 @default.
• W2946506871 cites W2135497861 @default.
• W2946506871 cites W2166816543 @default.
• W2946506871 cites W2183341477 @default.
• W2946506871 cites W2194775991 @default.
• W2946506871 cites W2286365479 @default.
• W2946506871 cites W2300242332 @default.
• W2946506871 cites W2524428287 @default.
• W2946506871 cites W2586654419 @default.
• W2946506871 cites W2606722458 @default.
• W2946506871 cites W2627042741 @default.
• W2946506871 cites W2783873922 @default.
• W2946506871 cites W2786771851 @default.
• W2946506871 cites W2787697768 @default.
• W2946506871 cites W2798544842 @default.
• W2946506871 cites W2798693909 @default.
• W2946506871 cites W2884150179 @default.
• W2946506871 cites W2886847195 @default.
• W2946506871 cites W2897049076 @default.
• W2946506871 cites W2904640605 @default.
• W2946506871 cites W2907544761 @default.
• W2946506871 cites W2926546867 @default.
• W2946506871 cites W2927163560 @default.
• W2946506871 cites W2929535330 @default.
• W2946506871 cites W2933062780 @default.
• W2946506871 cites W2934918255 @default.
• W2946506871 cites W2944508492 @default.
• W2946506871 cites W2949335192 @default.
• W2946506871 cites W2950726407 @default.
• W2946506871 cites W2962684187 @default.
• W2946506871 cites W2963163009 @default.
• W2946506871 cites W2963374099 @default.
• W2946506871 cites W2963701245 @default.
• W2946506871 cites W2963828549 @default.
• W2946506871 cites W2964299589 @default.
• W2946506871 cites W2966206672 @default.
• W2946506871 cites W2998218113 @default.
• W2946506871 hasPublicationYear "2019" @default.
• W2946506871 type Work @default.
• W2946506871 sameAs 2946506871 @default.
• W2946506871 citedByCount "2" @default.
• W2946506871 countsByYear W29465068712019 @default.
• W2946506871 countsByYear W29465068712020 @default.
• W2946506871 crossrefType "posted-content" @default.
• W2946506871 hasAuthorship W2946506871A5017964819 @default.
• W2946506871 hasAuthorship W2946506871A5019913171 @default.
• W2946506871 hasAuthorship W2946506871A5023269428 @default.
• W2946506871 hasAuthorship W2946506871A5025410418 @default.
• W2946506871 hasAuthorship W2946506871A5029515956 @default.
• W2946506871 hasAuthorship W2946506871A5031600563 @default.
• W2946506871 hasAuthorship W2946506871A5089135250 @default.
• W2946506871 hasAuthorship W2946506871A5091675091 @default.
• W2946506871 hasConcept C105795698 @default.
• W2946506871 hasConcept C111919701 @default.
• W2946506871 hasConcept C153180895 @default.
• W2946506871 hasConcept C154945302 @default.
• W2946506871 hasConcept C179518139 @default.
• W2946506871 hasConcept C2776214188 @default.
• W2946506871 hasConcept C33923547 @default.
• W2946506871 hasConcept C41008148 @default.
• W2946506871 hasConcept C74912251 @default.
• W2946506871 hasConcept C81363708 @default.
• W2946506871 hasConceptScore W2946506871C105795698 @default.
• W2946506871 hasConceptScore W2946506871C111919701 @default.
• W2946506871 hasConceptScore W2946506871C153180895 @default.
• W2946506871 hasConceptScore W2946506871C154945302 @default.
• W2946506871 hasConceptScore W2946506871C179518139 @default.
• W2946506871 hasConceptScore W2946506871C2776214188 @default.
• W2946506871 hasConceptScore W2946506871C33923547 @default.
• W2946506871 hasConceptScore W2946506871C41008148 @default.
• W2946506871 hasConceptScore W2946506871C74912251 @default.
• W2946506871 hasConceptScore W2946506871C81363708 @default.
• W2946506871 hasLocation W29465068711 @default.
• W2946506871 hasOpenAccess W2946506871 @default.
• W2946506871 hasPrimaryLocation W29465068711 @default.
• W2946506871 hasRelatedWork W2622872848 @default.
• W2946506871 hasRelatedWork W2801982910 @default.
• W2946506871 hasRelatedWork W2862715393 @default.
• W2946506871 hasRelatedWork W2907757276 @default.
• W2946506871 hasRelatedWork W2918006600 @default.
• W2946506871 hasRelatedWork W2950000202 @default.
• W2946506871 hasRelatedWork W2968754051 @default.
• W2946506871 hasRelatedWork W2973613048 @default.
• W2946506871 hasRelatedWork W2989199275 @default.
• W2946506871 hasRelatedWork W3004807204 @default.
• W2946506871 hasRelatedWork W3091520591 @default.

• next