Matches in SemOpenAlex for { <https://semopenalex.org/work/W3197256510> ?p ?o ?g. }
- W3197256510 endingPage "26" @default.
- W3197256510 startingPage "1" @default.
- W3197256510 abstract "Training deep learning networks involves continuous weight updates across the various layers of the deep network while using a backpropagation (BP) algorithm. This results in expensive computation overheads during training. Consequently, most deep learning accelerators today employ pretrained weights and focus only on improving the design of the inference phase. The recent trend is to build a complete deep learning accelerator by incorporating the training module. Such efforts require an ultra-fast chip architecture for executing the BP algorithm. In this article, we propose a novel photonics-based backpropagation accelerator for high-performance deep learning training. We present the design for a convolutional neural network (CNN), BPLight-CNN , which incorporates the silicon photonics-based backpropagation accelerator. BPLight-CNN is a first-of-its-kind photonic and memristor-based CNN architecture for end-to-end training and prediction. We evaluate BPLight-CNN using a photonic CAD framework (IPKISS) on deep learning benchmark models, including LeNet and VGG-Net. The proposed design achieves (i) at least 34× speedup, 34× improvement in computational efficiency, and 38.5× energy savings during training; and (ii) 29× speedup, 31× improvement in computational efficiency, and 38.7× improvement in energy savings during inference compared with the state-of-the-art designs. All of these comparisons are done at a 16-bit resolution, and BPLight-CNN achieves these improvements at a cost of approximately 6% lower accuracy compared with the state-of-the-art." @default.
- W3197256510 created "2021-09-13" @default.
- W3197256510 creator A5018382547 @default.
- W3197256510 creator A5041633893 @default.
- W3197256510 creator A5044935043 @default.
- W3197256510 creator A5057324688 @default.
- W3197256510 creator A5076611650 @default.
- W3197256510 date "2021-09-01" @default.
- W3197256510 modified "2023-10-16" @default.
- W3197256510 title "BPLight-CNN: A Photonics-Based Backpropagation Accelerator for Deep Learning" @default.
- W3197256510 cites W1981204644 @default.
- W3197256510 cites W1985818188 @default.
- W3197256510 cites W2013626493 @default.
- W3197256510 cites W2018774711 @default.
- W3197256510 cites W2053675542 @default.
- W3197256510 cites W2078107154 @default.
- W3197256510 cites W2091382331 @default.
- W3197256510 cites W2112181056 @default.
- W3197256510 cites W2112796928 @default.
- W3197256510 cites W2117539524 @default.
- W3197256510 cites W2126273023 @default.
- W3197256510 cites W2155893237 @default.
- W3197256510 cites W2164819109 @default.
- W3197256510 cites W2307193480 @default.
- W3197256510 cites W2399958287 @default.
- W3197256510 cites W2517214291 @default.
- W3197256510 cites W2518281301 @default.
- W3197256510 cites W2520083297 @default.
- W3197256510 cites W2527744069 @default.
- W3197256510 cites W2613989746 @default.
- W3197256510 cites W2618530766 @default.
- W3197256510 cites W2752849906 @default.
- W3197256510 cites W2783157379 @default.
- W3197256510 cites W2802266077 @default.
- W3197256510 cites W2805862866 @default.
- W3197256510 cites W2807750997 @default.
- W3197256510 cites W2898660405 @default.
- W3197256510 cites W2963387357 @default.
- W3197256510 cites W3012645150 @default.
- W3197256510 cites W3047623602 @default.
- W3197256510 cites W4248430148 @default.
- W3197256510 doi "https://doi.org/10.1145/3446212" @default.
- W3197256510 hasPublicationYear "2021" @default.
- W3197256510 type Work @default.
- W3197256510 sameAs 3197256510 @default.
- W3197256510 citedByCount "8" @default.
- W3197256510 countsByYear W31972565102021 @default.
- W3197256510 countsByYear W31972565102022 @default.
- W3197256510 countsByYear W31972565102023 @default.
- W3197256510 crossrefType "journal-article" @default.
- W3197256510 hasAuthorship W3197256510A5018382547 @default.
- W3197256510 hasAuthorship W3197256510A5041633893 @default.
- W3197256510 hasAuthorship W3197256510A5044935043 @default.
- W3197256510 hasAuthorship W3197256510A5057324688 @default.
- W3197256510 hasAuthorship W3197256510A5076611650 @default.
- W3197256510 hasBestOaLocation W31972565102 @default.
- W3197256510 hasConcept C108583219 @default.
- W3197256510 hasConcept C113775141 @default.
- W3197256510 hasConcept C119857082 @default.
- W3197256510 hasConcept C120665830 @default.
- W3197256510 hasConcept C121332964 @default.
- W3197256510 hasConcept C13280743 @default.
- W3197256510 hasConcept C154945302 @default.
- W3197256510 hasConcept C155032097 @default.
- W3197256510 hasConcept C173608175 @default.
- W3197256510 hasConcept C185798385 @default.
- W3197256510 hasConcept C205649164 @default.
- W3197256510 hasConcept C20788544 @default.
- W3197256510 hasConcept C2776214188 @default.
- W3197256510 hasConcept C41008148 @default.
- W3197256510 hasConcept C50644808 @default.
- W3197256510 hasConcept C68339613 @default.
- W3197256510 hasConcept C81363708 @default.
- W3197256510 hasConceptScore W3197256510C108583219 @default.
- W3197256510 hasConceptScore W3197256510C113775141 @default.
- W3197256510 hasConceptScore W3197256510C119857082 @default.
- W3197256510 hasConceptScore W3197256510C120665830 @default.
- W3197256510 hasConceptScore W3197256510C121332964 @default.
- W3197256510 hasConceptScore W3197256510C13280743 @default.
- W3197256510 hasConceptScore W3197256510C154945302 @default.
- W3197256510 hasConceptScore W3197256510C155032097 @default.
- W3197256510 hasConceptScore W3197256510C173608175 @default.
- W3197256510 hasConceptScore W3197256510C185798385 @default.
- W3197256510 hasConceptScore W3197256510C205649164 @default.
- W3197256510 hasConceptScore W3197256510C20788544 @default.
- W3197256510 hasConceptScore W3197256510C2776214188 @default.
- W3197256510 hasConceptScore W3197256510C41008148 @default.
- W3197256510 hasConceptScore W3197256510C50644808 @default.
- W3197256510 hasConceptScore W3197256510C68339613 @default.
- W3197256510 hasConceptScore W3197256510C81363708 @default.
- W3197256510 hasFunder F4320307136 @default.
- W3197256510 hasIssue "4" @default.
- W3197256510 hasLocation W31972565101 @default.
- W3197256510 hasLocation W31972565102 @default.
- W3197256510 hasLocation W31972565103 @default.
- W3197256510 hasOpenAccess W3197256510 @default.
- W3197256510 hasPrimaryLocation W31972565101 @default.
- W3197256510 hasRelatedWork W2731899572 @default.