FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2971381503> ?p ?o ?g. }

• W2971381503 endingPage "249" @default.
• W2971381503 startingPage "237" @default.
• W2971381503 abstract "Convolutional neural networks (CNNs) outperform traditional machine learning algorithms across a wide range of applications, such as object recognition, image segmentation, and autonomous driving. However, their ever-growing computational complexity makes it necessary to design efficient hardware accelerators. Most CNN accelerators focus on exploring various dataflow styles and designs that exploit computational parallelism. However, potential performance improvement (speedup) from sparsity has not been adequately addressed. The computation and memory footprint of CNNs can be significantly reduced if sparsity is exploited in network evaluations. To further improve performance and energy efficiency, some accelerators evaluate CNNs with limited precision. However, this is limited to the inference phase since reduced precision sacrifices network accuracy if used in training. In addition, CNN evaluation is usually memory-intensive, especially during training. The performance bottleneck arises from the fact that the memory cannot feed the computational units enough data, resulting in idling of these computational units and thus low utilization ratios. In this article, we propose SPRING, a SParsity-aware Reduced-precision Monolithic 3D CNN accelerator for trainING and inference. SPRING supports both CNN training and inference. It uses a binary mask scheme to encode sparsities in activations and weights. It uses the stochastic rounding algorithm to train CNNs with reduced precision without accuracy loss. To alleviate the memory bottleneck in CNN evaluation, especially during training, SPRING uses an efficient monolithic 3D nonvolatile memory interface to increase memory bandwidth. Compared to Nvidia GeForce GTX 1080 Ti, SPRING achieves 15.6×, 4.2×, and 66.0× improvements in performance, power reduction, and energy efficiency, respectively, for CNN training, and 15.5×, 4.5×, and 69.1× improvements in performance, power reduction, and energy efficiency, respectively, for inference." @default.
• W2971381503 created "2019-09-12" @default.
• W2971381503 creator A5044339131 @default.
• W2971381503 creator A5086131079 @default.
• W2971381503 date "2022-01-01" @default.
• W2971381503 modified "2023-09-27" @default.
• W2971381503 title "SPRING: A Sparsity-Aware Reduced-Precision Monolithic 3D CNN Accelerator Architecture for Training and Inference" @default.
• W2971381503 cites W1989931041 @default.
• W2971381503 cites W2010202670 @default.
• W2971381503 cites W2041811089 @default.
• W2971381503 cites W2048266589 @default.
• W2971381503 cites W2106033855 @default.
• W2971381503 cites W2117539524 @default.
• W2971381503 cites W2138162504 @default.
• W2971381503 cites W2152839228 @default.
• W2971381503 cites W2183341477 @default.
• W2971381503 cites W2204664557 @default.
• W2971381503 cites W2285660444 @default.
• W2971381503 cites W2442974303 @default.
• W2971381503 cites W2516141709 @default.
• W2971381503 cites W2541839172 @default.
• W2971381503 cites W2605347906 @default.
• W2971381503 cites W2606722458 @default.
• W2971381503 cites W2612076670 @default.
• W2971381503 cites W2625457103 @default.
• W2971381503 cites W2880869856 @default.
• W2971381503 cites W2900749507 @default.
• W2971381503 cites W2904902077 @default.
• W2971381503 cites W2913716762 @default.
• W2971381503 cites W2949866178 @default.
• W2971381503 cites W2962821792 @default.
• W2971381503 cites W2963163009 @default.
• W2971381503 cites W2963594949 @default.
• W2971381503 cites W2963821229 @default.
• W2971381503 cites W2964081807 @default.
• W2971381503 cites W2964350391 @default.
• W2971381503 cites W3103034165 @default.
• W2971381503 cites W3104393472 @default.
• W2971381503 cites W4240168186 @default.
• W2971381503 cites W4242765802 @default.
• W2971381503 cites W4247198796 @default.
• W2971381503 cites W4249932213 @default.
• W2971381503 cites W4256629673 @default.
• W2971381503 doi "https://doi.org/10.1109/tetc.2020.3003328" @default.
• W2971381503 hasPublicationYear "2022" @default.
• W2971381503 type Work @default.
• W2971381503 sameAs 2971381503 @default.
• W2971381503 citedByCount "7" @default.
• W2971381503 countsByYear W29713815032020 @default.
• W2971381503 countsByYear W29713815032021 @default.
• W2971381503 countsByYear W29713815032022 @default.
• W2971381503 countsByYear W29713815032023 @default.
• W2971381503 crossrefType "journal-article" @default.
• W2971381503 hasAuthorship W2971381503A5044339131 @default.
• W2971381503 hasAuthorship W2971381503A5086131079 @default.
• W2971381503 hasBestOaLocation W29713815032 @default.
• W2971381503 hasConcept C108583219 @default.
• W2971381503 hasConcept C111919701 @default.
• W2971381503 hasConcept C113775141 @default.
• W2971381503 hasConcept C11413529 @default.
• W2971381503 hasConcept C119599485 @default.
• W2971381503 hasConcept C127413603 @default.
• W2971381503 hasConcept C149635348 @default.
• W2971381503 hasConcept C154945302 @default.
• W2971381503 hasConcept C173608175 @default.
• W2971381503 hasConcept C179799912 @default.
• W2971381503 hasConcept C188045654 @default.
• W2971381503 hasConcept C2742236 @default.
• W2971381503 hasConcept C2776214188 @default.
• W2971381503 hasConcept C2780513914 @default.
• W2971381503 hasConcept C41008148 @default.
• W2971381503 hasConcept C68339613 @default.
• W2971381503 hasConcept C74912251 @default.
• W2971381503 hasConcept C81363708 @default.
• W2971381503 hasConcept C84211073 @default.
• W2971381503 hasConcept C96324660 @default.
• W2971381503 hasConceptScore W2971381503C108583219 @default.
• W2971381503 hasConceptScore W2971381503C111919701 @default.
• W2971381503 hasConceptScore W2971381503C113775141 @default.
• W2971381503 hasConceptScore W2971381503C11413529 @default.
• W2971381503 hasConceptScore W2971381503C119599485 @default.
• W2971381503 hasConceptScore W2971381503C127413603 @default.
• W2971381503 hasConceptScore W2971381503C149635348 @default.
• W2971381503 hasConceptScore W2971381503C154945302 @default.
• W2971381503 hasConceptScore W2971381503C173608175 @default.
• W2971381503 hasConceptScore W2971381503C179799912 @default.
• W2971381503 hasConceptScore W2971381503C188045654 @default.
• W2971381503 hasConceptScore W2971381503C2742236 @default.
• W2971381503 hasConceptScore W2971381503C2776214188 @default.
• W2971381503 hasConceptScore W2971381503C2780513914 @default.
• W2971381503 hasConceptScore W2971381503C41008148 @default.
• W2971381503 hasConceptScore W2971381503C68339613 @default.
• W2971381503 hasConceptScore W2971381503C74912251 @default.
• W2971381503 hasConceptScore W2971381503C81363708 @default.
• W2971381503 hasConceptScore W2971381503C84211073 @default.
• W2971381503 hasConceptScore W2971381503C96324660 @default.
• W2971381503 hasFunder F4320306076 @default.
• W2971381503 hasIssue "1" @default.

• next