FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387002168> ?p ?o ?g. }

• W4387002168 abstract "The constant drive to achieve higher performance in deep neural networks (DNNs) has led to the proliferation of very large models. Model training, however, requires intensive computation time and energy. Memristor-based compute-in-memory (CIM) modules can perform vector-matrix multiplication (VMM) in place and in parallel, and have shown great promises in DNN inference applications. However, CIM-based model training faces challenges due to non-linear weight updates, device variations, and low-precision. In this work, a mixed-precision training scheme is experimentally implemented to mitigate these effects using a bulk-switching memristor-based CIM module. Low-precision CIM modules are used to accelerate the expensive VMM operations, with high-precision weight updates accumulated in digital units. Memristor devices are only changed when the accumulated weight update value exceeds a pre-defined threshold. The proposed scheme is implemented with a system-onchip of fully integrated analog CIM modules and digital sub-systems, showing fast convergence of LeNet training to 97.73%. The efficacy of training larger models is evaluated using realistic hardware parameters and verifies that CIM modules can enable efficient mix-precision DNN training with accuracy comparable to full-precision software-trained models. Additionally, models trained on chip are inherently robust to hardware variations, allowing direct mapping to CIM inference chips without additional re-training." @default.
• W4387002168 created "2023-09-26" @default.
• W4387002168 creator A5021229566 @default.
• W4387002168 creator A5031961057 @default.
• W4387002168 creator A5054528638 @default.
• W4387002168 creator A5055814385 @default.
• W4387002168 creator A5056081092 @default.
• W4387002168 creator A5058990485 @default.
• W4387002168 creator A5064627383 @default.
• W4387002168 creator A5066881061 @default.
• W4387002168 date "2023-10-15" @default.
• W4387002168 modified "2023-10-18" @default.
• W4387002168 title "Bulk‐Switching Memristor‐Based Compute‐In‐Memory Module for Deep Neural Network Training" @default.
• W4387002168 cites W1978704421 @default.
• W4387002168 cites W1983964264 @default.
• W4387002168 cites W2036899386 @default.
• W4387002168 cites W2043707274 @default.
• W4387002168 cites W2079113849 @default.
• W4387002168 cites W2118980095 @default.
• W4387002168 cites W2300242332 @default.
• W4387002168 cites W2468323742 @default.
• W4387002168 cites W2526646482 @default.
• W4387002168 cites W2591029953 @default.
• W4387002168 cites W2762731122 @default.
• W4387002168 cites W2765081478 @default.
• W4387002168 cites W2803163155 @default.
• W4387002168 cites W2923010225 @default.
• W4387002168 cites W2942216650 @default.
• W4387002168 cites W2945187116 @default.
• W4387002168 cites W2960778947 @default.
• W4387002168 cites W3003821665 @default.
• W4387002168 cites W3007317489 @default.
• W4387002168 cites W3025017204 @default.
• W4387002168 cites W3091018873 @default.
• W4387002168 cites W3101272433 @default.
• W4387002168 cites W3177828909 @default.
• W4387002168 cites W3206839852 @default.
• W4387002168 cites W4200398297 @default.
• W4387002168 cites W4205937980 @default.
• W4387002168 cites W4213193842 @default.
• W4387002168 cites W4225278450 @default.
• W4387002168 cites W4292121737 @default.
• W4387002168 cites W4312131723 @default.
• W4387002168 doi "https://doi.org/10.1002/adma.202305465" @default.
• W4387002168 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37747134" @default.
• W4387002168 hasPublicationYear "2023" @default.
• W4387002168 type Work @default.
• W4387002168 citedByCount "0" @default.
• W4387002168 crossrefType "journal-article" @default.
• W4387002168 hasAuthorship W4387002168A5021229566 @default.
• W4387002168 hasAuthorship W4387002168A5031961057 @default.
• W4387002168 hasAuthorship W4387002168A5054528638 @default.
• W4387002168 hasAuthorship W4387002168A5055814385 @default.
• W4387002168 hasAuthorship W4387002168A5056081092 @default.
• W4387002168 hasAuthorship W4387002168A5058990485 @default.
• W4387002168 hasAuthorship W4387002168A5064627383 @default.
• W4387002168 hasAuthorship W4387002168A5066881061 @default.
• W4387002168 hasBestOaLocation W43870021681 @default.
• W4387002168 hasConcept C113775141 @default.
• W4387002168 hasConcept C11413529 @default.
• W4387002168 hasConcept C127413603 @default.
• W4387002168 hasConcept C134306372 @default.
• W4387002168 hasConcept C150072547 @default.
• W4387002168 hasConcept C154945302 @default.
• W4387002168 hasConcept C24326235 @default.
• W4387002168 hasConcept C2776214188 @default.
• W4387002168 hasConcept C33923547 @default.
• W4387002168 hasConcept C41008148 @default.
• W4387002168 hasConcept C45374587 @default.
• W4387002168 hasConcept C50644808 @default.
• W4387002168 hasConcept C77618280 @default.
• W4387002168 hasConcept C9390403 @default.
• W4387002168 hasConceptScore W4387002168C113775141 @default.
• W4387002168 hasConceptScore W4387002168C11413529 @default.
• W4387002168 hasConceptScore W4387002168C127413603 @default.
• W4387002168 hasConceptScore W4387002168C134306372 @default.
• W4387002168 hasConceptScore W4387002168C150072547 @default.
• W4387002168 hasConceptScore W4387002168C154945302 @default.
• W4387002168 hasConceptScore W4387002168C24326235 @default.
• W4387002168 hasConceptScore W4387002168C2776214188 @default.
• W4387002168 hasConceptScore W4387002168C33923547 @default.
• W4387002168 hasConceptScore W4387002168C41008148 @default.
• W4387002168 hasConceptScore W4387002168C45374587 @default.
• W4387002168 hasConceptScore W4387002168C50644808 @default.
• W4387002168 hasConceptScore W4387002168C77618280 @default.
• W4387002168 hasConceptScore W4387002168C9390403 @default.
• W4387002168 hasFunder F4320306076 @default.
• W4387002168 hasLocation W43870021681 @default.
• W4387002168 hasLocation W43870021682 @default.
• W4387002168 hasOpenAccess W4387002168 @default.
• W4387002168 hasPrimaryLocation W43870021681 @default.
• W4387002168 hasRelatedWork W2052332160 @default.
• W4387002168 hasRelatedWork W2204001882 @default.
• W4387002168 hasRelatedWork W2264938185 @default.
• W4387002168 hasRelatedWork W2304829496 @default.
• W4387002168 hasRelatedWork W2358307108 @default.
• W4387002168 hasRelatedWork W2463286374 @default.
• W4387002168 hasRelatedWork W2548107569 @default.
• W4387002168 hasRelatedWork W2966276069 @default.
• W4387002168 hasRelatedWork W3031124155 @default.

• next