Matches in SemOpenAlex for { <https://semopenalex.org/work/W3135280793> ?p ?o ?g. }
Showing items 1 to 100 of
100
with 100 items per page.
- W3135280793 abstract "Deep learning has been successful in many fields such as acoustics, image, and natural language processing. However, due to the unique characteristics of graphs, deep learning using universal graph data is not easy. The Graph Attention Networks (GATs) show the best performance in multiple authoritative node classification benchmark tests (including transductive and inductive). The purpose of this research is to design and implement an FPGA-based accelerator called S-GAT for graph attention networks that achieves excellent performance on acceleration and energy efficiency without losing accuracy, and does not rely on DSPs and large amounts of on-chip memory. We design S-GAT with software and hardware co-optimization. Specifically, we use model compression and feature quantization to reduce the model size, and use shift addition units (SAUs) to convert multiplication into shift operation to further reduce the computation requirements. We integrate the above optimizations into a universal hardware pipeline for various structures of GATs. At last, we evaluate our design on an Inspur F10A board with an Intel Arria 10 GX1150 and 16 GB DDR3 memory. Experimental results show that S-GAT can achieve 7.34 times speedup over Nvidia Tesla V100 and 593 times over Xeon CPU Gold 5115 while maintaining accuracy, and 48 times and 2400 times on energy efficiency respectively." @default.
- W3135280793 created "2021-03-15" @default.
- W3135280793 creator A5005346872 @default.
- W3135280793 creator A5049265630 @default.
- W3135280793 creator A5070216219 @default.
- W3135280793 date "2020-12-01" @default.
- W3135280793 modified "2023-09-27" @default.
- W3135280793 title "S-GAT: Accelerating Graph Attention Networks Inference on FPGA Platform with Shift Operation" @default.
- W3135280793 cites W2294282016 @default.
- W3135280793 cites W2519887557 @default.
- W3135280793 cites W2584785188 @default.
- W3135280793 cites W2785456876 @default.
- W3135280793 cites W2788512147 @default.
- W3135280793 cites W2788919350 @default.
- W3135280793 cites W2789007918 @default.
- W3135280793 cites W2796276162 @default.
- W3135280793 cites W2891335318 @default.
- W3135280793 cites W2907463061 @default.
- W3135280793 cites W2910396952 @default.
- W3135280793 cites W2935524202 @default.
- W3135280793 cites W2946480111 @default.
- W3135280793 cites W2952202138 @default.
- W3135280793 cites W2963858333 @default.
- W3135280793 cites W2965469401 @default.
- W3135280793 cites W3004884952 @default.
- W3135280793 cites W3010314017 @default.
- W3135280793 cites W3011667710 @default.
- W3135280793 cites W3017228913 @default.
- W3135280793 cites W3105753905 @default.
- W3135280793 cites W3106125969 @default.
- W3135280793 doi "https://doi.org/10.1109/icpads51040.2020.00093" @default.
- W3135280793 hasPublicationYear "2020" @default.
- W3135280793 type Work @default.
- W3135280793 sameAs 3135280793 @default.
- W3135280793 citedByCount "0" @default.
- W3135280793 crossrefType "proceedings-article" @default.
- W3135280793 hasAuthorship W3135280793A5005346872 @default.
- W3135280793 hasAuthorship W3135280793A5049265630 @default.
- W3135280793 hasAuthorship W3135280793A5070216219 @default.
- W3135280793 hasConcept C108583219 @default.
- W3135280793 hasConcept C113775141 @default.
- W3135280793 hasConcept C11413529 @default.
- W3135280793 hasConcept C13164978 @default.
- W3135280793 hasConcept C132525143 @default.
- W3135280793 hasConcept C13280743 @default.
- W3135280793 hasConcept C145108525 @default.
- W3135280793 hasConcept C149635348 @default.
- W3135280793 hasConcept C154945302 @default.
- W3135280793 hasConcept C169590947 @default.
- W3135280793 hasConcept C173608175 @default.
- W3135280793 hasConcept C185798385 @default.
- W3135280793 hasConcept C199360897 @default.
- W3135280793 hasConcept C205649164 @default.
- W3135280793 hasConcept C2776214188 @default.
- W3135280793 hasConcept C28855332 @default.
- W3135280793 hasConcept C41008148 @default.
- W3135280793 hasConcept C42935608 @default.
- W3135280793 hasConcept C45374587 @default.
- W3135280793 hasConcept C68339613 @default.
- W3135280793 hasConcept C76970557 @default.
- W3135280793 hasConcept C80444323 @default.
- W3135280793 hasConceptScore W3135280793C108583219 @default.
- W3135280793 hasConceptScore W3135280793C113775141 @default.
- W3135280793 hasConceptScore W3135280793C11413529 @default.
- W3135280793 hasConceptScore W3135280793C13164978 @default.
- W3135280793 hasConceptScore W3135280793C132525143 @default.
- W3135280793 hasConceptScore W3135280793C13280743 @default.
- W3135280793 hasConceptScore W3135280793C145108525 @default.
- W3135280793 hasConceptScore W3135280793C149635348 @default.
- W3135280793 hasConceptScore W3135280793C154945302 @default.
- W3135280793 hasConceptScore W3135280793C169590947 @default.
- W3135280793 hasConceptScore W3135280793C173608175 @default.
- W3135280793 hasConceptScore W3135280793C185798385 @default.
- W3135280793 hasConceptScore W3135280793C199360897 @default.
- W3135280793 hasConceptScore W3135280793C205649164 @default.
- W3135280793 hasConceptScore W3135280793C2776214188 @default.
- W3135280793 hasConceptScore W3135280793C28855332 @default.
- W3135280793 hasConceptScore W3135280793C41008148 @default.
- W3135280793 hasConceptScore W3135280793C42935608 @default.
- W3135280793 hasConceptScore W3135280793C45374587 @default.
- W3135280793 hasConceptScore W3135280793C68339613 @default.
- W3135280793 hasConceptScore W3135280793C76970557 @default.
- W3135280793 hasConceptScore W3135280793C80444323 @default.
- W3135280793 hasLocation W31352807931 @default.
- W3135280793 hasOpenAccess W3135280793 @default.
- W3135280793 hasPrimaryLocation W31352807931 @default.
- W3135280793 hasRelatedWork W1188123746 @default.
- W3135280793 hasRelatedWork W2053732522 @default.
- W3135280793 hasRelatedWork W2335592298 @default.
- W3135280793 hasRelatedWork W3021371744 @default.
- W3135280793 hasRelatedWork W3135280793 @default.
- W3135280793 hasRelatedWork W3216454443 @default.
- W3135280793 hasRelatedWork W4200123712 @default.
- W3135280793 hasRelatedWork W4214908790 @default.
- W3135280793 hasRelatedWork W4286233754 @default.
- W3135280793 hasRelatedWork W2093143612 @default.
- W3135280793 isParatext "false" @default.
- W3135280793 isRetracted "false" @default.
- W3135280793 magId "3135280793" @default.
- W3135280793 workType "article" @default.