FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3136309901> ?p ?o ?g. }

• W3136309901 endingPage "2463" @default.
• W3136309901 startingPage "2448" @default.
• W3136309901 abstract "Deep Neural Networks (DNNs) have emerged as an important class of machine learning algorithms, providing accurate solutions to a broad range of applications. Sparsity in activation maps in DNN training presents an opportunity to reduce computations. However, exploiting activation sparsity presents two major challenges: i) profiling activation sparsity during training comes with significant overhead due to computing the degree of sparsity and the data movement; ii) the dynamic nature of activation maps requires dynamic dense-to-sparse conversion during training, leading to significant overhead. In this article, we present Spartan, a lightweight hardware/software framework to accelerate DNN training on a GPU. Spartan provides a cost-effective and programmer-transparent microarchitectural solution to exploit activation sparsity detected during training. Spartan provides an efficient sparsity monitor, a tile-based sparse GEMM algorithm, and a novel compaction engine designed for GPU workloads. Spartan can reduce sparsity profiling overhead by 52.5× on average. For the most compute-intensive layers, i.e., convolutional layers, we can speedup AlexNet by 3.4×, VGGNet-16 by 2.14×, and ResNet-18 by 2.02×, when training on the ImageNet dataset." @default.
• W3136309901 created "2021-03-29" @default.
• W3136309901 creator A5007832959 @default.
• W3136309901 creator A5008501980 @default.
• W3136309901 creator A5022915639 @default.
• W3136309901 creator A5041025969 @default.
• W3136309901 creator A5057498242 @default.
• W3136309901 creator A5058119214 @default.
• W3136309901 creator A5061128237 @default.
• W3136309901 creator A5066123777 @default.
• W3136309901 creator A5080113933 @default.
• W3136309901 date "2021-10-01" @default.
• W3136309901 modified "2023-10-14" @default.
• W3136309901 title "Spartan: A Sparsity-Adaptive Framework to Accelerate Deep Neural Network Training on GPUs" @default.
• W3136309901 cites W2007259720 @default.
• W3136309901 cites W2097117768 @default.
• W3136309901 cites W2108598243 @default.
• W3136309901 cites W2142801765 @default.
• W3136309901 cites W2194775991 @default.
• W3136309901 cites W2285660444 @default.
• W3136309901 cites W2513554817 @default.
• W3136309901 cites W2516141709 @default.
• W3136309901 cites W2585720638 @default.
• W3136309901 cites W2587914027 @default.
• W3136309901 cites W2657126969 @default.
• W3136309901 cites W2795994638 @default.
• W3136309901 cites W2796013597 @default.
• W3136309901 cites W2883283076 @default.
• W3136309901 cites W2890392327 @default.
• W3136309901 cites W2904902077 @default.
• W3136309901 cites W2915106038 @default.
• W3136309901 cites W2945146780 @default.
• W3136309901 cites W2949674408 @default.
• W3136309901 cites W2952928793 @default.
• W3136309901 cites W2962821792 @default.
• W3136309901 cites W2963896595 @default.
• W3136309901 cites W2966206672 @default.
• W3136309901 cites W3012561096 @default.
• W3136309901 cites W3016542674 @default.
• W3136309901 cites W3104805932 @default.
• W3136309901 doi "https://doi.org/10.1109/tpds.2021.3067825" @default.
• W3136309901 hasPublicationYear "2021" @default.
• W3136309901 type Work @default.
• W3136309901 sameAs 3136309901 @default.
• W3136309901 citedByCount "3" @default.
• W3136309901 countsByYear W31363099012022 @default.
• W3136309901 countsByYear W31363099012023 @default.
• W3136309901 crossrefType "journal-article" @default.
• W3136309901 hasAuthorship W3136309901A5007832959 @default.
• W3136309901 hasAuthorship W3136309901A5008501980 @default.
• W3136309901 hasAuthorship W3136309901A5022915639 @default.
• W3136309901 hasAuthorship W3136309901A5041025969 @default.
• W3136309901 hasAuthorship W3136309901A5057498242 @default.
• W3136309901 hasAuthorship W3136309901A5058119214 @default.
• W3136309901 hasAuthorship W3136309901A5061128237 @default.
• W3136309901 hasAuthorship W3136309901A5066123777 @default.
• W3136309901 hasAuthorship W3136309901A5080113933 @default.
• W3136309901 hasBestOaLocation W31363099012 @default.
• W3136309901 hasConcept C10689553 @default.
• W3136309901 hasConcept C108583219 @default.
• W3136309901 hasConcept C111919701 @default.
• W3136309901 hasConcept C118524514 @default.
• W3136309901 hasConcept C121332964 @default.
• W3136309901 hasConcept C149635348 @default.
• W3136309901 hasConcept C154945302 @default.
• W3136309901 hasConcept C163716315 @default.
• W3136309901 hasConcept C165696696 @default.
• W3136309901 hasConcept C173608175 @default.
• W3136309901 hasConcept C187191949 @default.
• W3136309901 hasConcept C2778514511 @default.
• W3136309901 hasConcept C2779960059 @default.
• W3136309901 hasConcept C2984842247 @default.
• W3136309901 hasConcept C38652104 @default.
• W3136309901 hasConcept C41008148 @default.
• W3136309901 hasConcept C42935608 @default.
• W3136309901 hasConcept C50644808 @default.
• W3136309901 hasConcept C56372850 @default.
• W3136309901 hasConcept C62520636 @default.
• W3136309901 hasConcept C68339613 @default.
• W3136309901 hasConcept C81363708 @default.
• W3136309901 hasConceptScore W3136309901C10689553 @default.
• W3136309901 hasConceptScore W3136309901C108583219 @default.
• W3136309901 hasConceptScore W3136309901C111919701 @default.
• W3136309901 hasConceptScore W3136309901C118524514 @default.
• W3136309901 hasConceptScore W3136309901C121332964 @default.
• W3136309901 hasConceptScore W3136309901C149635348 @default.
• W3136309901 hasConceptScore W3136309901C154945302 @default.
• W3136309901 hasConceptScore W3136309901C163716315 @default.
• W3136309901 hasConceptScore W3136309901C165696696 @default.
• W3136309901 hasConceptScore W3136309901C173608175 @default.
• W3136309901 hasConceptScore W3136309901C187191949 @default.
• W3136309901 hasConceptScore W3136309901C2778514511 @default.
• W3136309901 hasConceptScore W3136309901C2779960059 @default.
• W3136309901 hasConceptScore W3136309901C2984842247 @default.
• W3136309901 hasConceptScore W3136309901C38652104 @default.
• W3136309901 hasConceptScore W3136309901C41008148 @default.
• W3136309901 hasConceptScore W3136309901C42935608 @default.
• W3136309901 hasConceptScore W3136309901C50644808 @default.

• next