Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385270611> ?p ?o ?g. }
- W4385270611 abstract "Training deep learning (DL) models often takes a significant amount of time and is thus typically performed on expensive GPUs to speed up the process. However, data loading has recently been identified as one of the main performance bottlenecks in DL, resulting in GPU under-utilization. Looking forward, the combination of larger datasets and faster GPUs will exacerbate the problem. The data management community has started to address this by proposing data loading optimization techniques, including lossy image compression. While lossy compression is a conceptually promising approach for mitigating data loading bottlenecks in DL, there is only limited understanding of its efficacy in terms of impact on model throughput and accuracy. In this paper, we present an extensive experimental analysis of lossy image compression as a means to improve the performance of neural network training. We find that lossy compression can improve both throughput and accuracy of DL pipelines if resources such as time or storage capacity are limited. Furthermore, the choice of compression quality and codec are important hyperparameters when training deep neural networks." @default.
- W4385270611 created "2023-07-27" @default.
- W4385270611 creator A5002413906 @default.
- W4385270611 creator A5066004392 @default.
- W4385270611 creator A5083942725 @default.
- W4385270611 creator A5087402246 @default.
- W4385270611 creator A5092545429 @default.
- W4385270611 date "2023-04-01" @default.
- W4385270611 modified "2023-09-30" @default.
- W4385270611 title "The Art of Losing to Win: Using Lossy Image Compression to Improve Data Loading in Deep Learning Pipelines" @default.
- W4385270611 cites W1558125431 @default.
- W4385270611 cites W2117539524 @default.
- W4385270611 cites W2165698076 @default.
- W4385270611 cites W2194775991 @default.
- W4385270611 cites W2618530766 @default.
- W4385270611 cites W2732026016 @default.
- W4385270611 cites W2900042580 @default.
- W4385270611 cites W2909678128 @default.
- W4385270611 cites W2963149687 @default.
- W4385270611 cites W2963268689 @default.
- W4385270611 cites W2963372104 @default.
- W4385270611 cites W2963980515 @default.
- W4385270611 cites W2964330541 @default.
- W4385270611 cites W2982083293 @default.
- W4385270611 cites W2982316857 @default.
- W4385270611 cites W3008591352 @default.
- W4385270611 cites W3017549830 @default.
- W4385270611 cites W3034469748 @default.
- W4385270611 cites W3048346032 @default.
- W4385270611 cites W3081844508 @default.
- W4385270611 cites W3088546539 @default.
- W4385270611 cites W3105067168 @default.
- W4385270611 cites W3106772683 @default.
- W4385270611 cites W3138303811 @default.
- W4385270611 cites W3174472699 @default.
- W4385270611 cites W3175089547 @default.
- W4385270611 cites W3196732841 @default.
- W4385270611 cites W3198095237 @default.
- W4385270611 cites W4220938834 @default.
- W4385270611 cites W4281788184 @default.
- W4385270611 cites W4294106961 @default.
- W4385270611 doi "https://doi.org/10.1109/icde55515.2023.00077" @default.
- W4385270611 hasPublicationYear "2023" @default.
- W4385270611 type Work @default.
- W4385270611 citedByCount "0" @default.
- W4385270611 crossrefType "proceedings-article" @default.
- W4385270611 hasAuthorship W4385270611A5002413906 @default.
- W4385270611 hasAuthorship W4385270611A5066004392 @default.
- W4385270611 hasAuthorship W4385270611A5083942725 @default.
- W4385270611 hasAuthorship W4385270611A5087402246 @default.
- W4385270611 hasAuthorship W4385270611A5092545429 @default.
- W4385270611 hasConcept C108583219 @default.
- W4385270611 hasConcept C113775141 @default.
- W4385270611 hasConcept C115961682 @default.
- W4385270611 hasConcept C119857082 @default.
- W4385270611 hasConcept C13481523 @default.
- W4385270611 hasConcept C154945302 @default.
- W4385270611 hasConcept C157764524 @default.
- W4385270611 hasConcept C159985019 @default.
- W4385270611 hasConcept C161765866 @default.
- W4385270611 hasConcept C165021410 @default.
- W4385270611 hasConcept C180016635 @default.
- W4385270611 hasConcept C192562407 @default.
- W4385270611 hasConcept C41008148 @default.
- W4385270611 hasConcept C50644808 @default.
- W4385270611 hasConcept C555944384 @default.
- W4385270611 hasConcept C76155785 @default.
- W4385270611 hasConcept C78548338 @default.
- W4385270611 hasConcept C9390403 @default.
- W4385270611 hasConcept C9417928 @default.
- W4385270611 hasConceptScore W4385270611C108583219 @default.
- W4385270611 hasConceptScore W4385270611C113775141 @default.
- W4385270611 hasConceptScore W4385270611C115961682 @default.
- W4385270611 hasConceptScore W4385270611C119857082 @default.
- W4385270611 hasConceptScore W4385270611C13481523 @default.
- W4385270611 hasConceptScore W4385270611C154945302 @default.
- W4385270611 hasConceptScore W4385270611C157764524 @default.
- W4385270611 hasConceptScore W4385270611C159985019 @default.
- W4385270611 hasConceptScore W4385270611C161765866 @default.
- W4385270611 hasConceptScore W4385270611C165021410 @default.
- W4385270611 hasConceptScore W4385270611C180016635 @default.
- W4385270611 hasConceptScore W4385270611C192562407 @default.
- W4385270611 hasConceptScore W4385270611C41008148 @default.
- W4385270611 hasConceptScore W4385270611C50644808 @default.
- W4385270611 hasConceptScore W4385270611C555944384 @default.
- W4385270611 hasConceptScore W4385270611C76155785 @default.
- W4385270611 hasConceptScore W4385270611C78548338 @default.
- W4385270611 hasConceptScore W4385270611C9390403 @default.
- W4385270611 hasConceptScore W4385270611C9417928 @default.
- W4385270611 hasLocation W43852706111 @default.
- W4385270611 hasOpenAccess W4385270611 @default.
- W4385270611 hasPrimaryLocation W43852706111 @default.
- W4385270611 hasRelatedWork W1544606459 @default.
- W4385270611 hasRelatedWork W2032304918 @default.
- W4385270611 hasRelatedWork W2053539432 @default.
- W4385270611 hasRelatedWork W2082502322 @default.
- W4385270611 hasRelatedWork W2113522614 @default.
- W4385270611 hasRelatedWork W2372078508 @default.
- W4385270611 hasRelatedWork W2405432330 @default.
- W4385270611 hasRelatedWork W2942043928 @default.