Matches in SemOpenAlex for { <https://semopenalex.org/work/W4312121097> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W4312121097 abstract "Computing-in-Memory (CiM) architectures based on emerging non-volatile memory (NVM) devices have demonstrated great potential for deep neural network (DNN) acceleration thanks to their high energy efficiency. However, NVM devices suffer from various non-idealities, especially device-to-device variations due to fabrication defects and cycle-to-cycle variations due to the stochastic behavior of devices. As such, the DNN weights actually mapped to NVM devices could deviate significantly from the expected values, leading to large performance degradation. To address this issue, most existing works focus on maximizing average performance under device variations. This objective would work well for general-purpose scenarios. But for safety-critical applications, the worst-case performance must also be considered. Unfortunately, this has been rarely explored in the literature. In this work, we formulate the problem of determining the worst-case performance of CiM DNN accelerators under the impact of device variations. We further propose a method to effectively find the specific combination of device variation in the high-dimensional space that leads to the worst-case performance. We find that even with very small device variations, the accuracy of a DNN can drop drastically, causing concerns when deploying CiM accelerators in safety-critical applications. Finally, we show that surprisingly none of the existing methods used to enhance average DNN performance in CiM accelerators are very effective when extended to enhance the worst-case performance, and further research down the road is needed to address this problem." @default.
- W4312121097 created "2023-01-04" @default.
- W4312121097 creator A5000141831 @default.
- W4312121097 creator A5053986599 @default.
- W4312121097 creator A5059093077 @default.
- W4312121097 date "2022-10-30" @default.
- W4312121097 modified "2023-10-05" @default.
- W4312121097 title "Computing-In-Memory Neural Network Accelerators for Safety-Critical Systems" @default.
- W4312121097 cites W2007339694 @default.
- W4312121097 cites W2108598243 @default.
- W4312121097 cites W2112796928 @default.
- W4312121097 cites W2194775991 @default.
- W4312121097 cites W2442974303 @default.
- W4312121097 cites W2518281301 @default.
- W4312121097 cites W2604319603 @default.
- W4312121097 cites W2795127895 @default.
- W4312121097 cites W2945387900 @default.
- W4312121097 cites W2945639226 @default.
- W4312121097 cites W2946522000 @default.
- W4312121097 cites W2997581110 @default.
- W4312121097 cites W3013462594 @default.
- W4312121097 cites W3022053993 @default.
- W4312121097 cites W3085709740 @default.
- W4312121097 cites W3092334294 @default.
- W4312121097 cites W3177878517 @default.
- W4312121097 cites W3211203038 @default.
- W4312121097 cites W3211724015 @default.
- W4312121097 cites W3212633509 @default.
- W4312121097 cites W4200496545 @default.
- W4312121097 cites W4200557188 @default.
- W4312121097 cites W4226216109 @default.
- W4312121097 cites W4293023585 @default.
- W4312121097 cites W4293024037 @default.
- W4312121097 doi "https://doi.org/10.1145/3508352.3549360" @default.
- W4312121097 hasPublicationYear "2022" @default.
- W4312121097 type Work @default.
- W4312121097 citedByCount "1" @default.
- W4312121097 countsByYear W43121210972023 @default.
- W4312121097 crossrefType "proceedings-article" @default.
- W4312121097 hasAuthorship W4312121097A5000141831 @default.
- W4312121097 hasAuthorship W4312121097A5053986599 @default.
- W4312121097 hasAuthorship W4312121097A5059093077 @default.
- W4312121097 hasBestOaLocation W43121210971 @default.
- W4312121097 hasConcept C117896860 @default.
- W4312121097 hasConcept C120665830 @default.
- W4312121097 hasConcept C121332964 @default.
- W4312121097 hasConcept C149635348 @default.
- W4312121097 hasConcept C154945302 @default.
- W4312121097 hasConcept C192209626 @default.
- W4312121097 hasConcept C41008148 @default.
- W4312121097 hasConcept C50644808 @default.
- W4312121097 hasConcept C74650414 @default.
- W4312121097 hasConceptScore W4312121097C117896860 @default.
- W4312121097 hasConceptScore W4312121097C120665830 @default.
- W4312121097 hasConceptScore W4312121097C121332964 @default.
- W4312121097 hasConceptScore W4312121097C149635348 @default.
- W4312121097 hasConceptScore W4312121097C154945302 @default.
- W4312121097 hasConceptScore W4312121097C192209626 @default.
- W4312121097 hasConceptScore W4312121097C41008148 @default.
- W4312121097 hasConceptScore W4312121097C50644808 @default.
- W4312121097 hasConceptScore W4312121097C74650414 @default.
- W4312121097 hasFunder F4320306076 @default.
- W4312121097 hasLocation W43121210971 @default.
- W4312121097 hasLocation W43121210972 @default.
- W4312121097 hasLocation W43121210973 @default.
- W4312121097 hasOpenAccess W4312121097 @default.
- W4312121097 hasPrimaryLocation W43121210971 @default.
- W4312121097 hasRelatedWork W1584420410 @default.
- W4312121097 hasRelatedWork W2015192756 @default.
- W4312121097 hasRelatedWork W2022840179 @default.
- W4312121097 hasRelatedWork W2120815133 @default.
- W4312121097 hasRelatedWork W2374512474 @default.
- W4312121097 hasRelatedWork W2386387936 @default.
- W4312121097 hasRelatedWork W2544032439 @default.
- W4312121097 hasRelatedWork W2909853096 @default.
- W4312121097 hasRelatedWork W4249271267 @default.
- W4312121097 hasRelatedWork W4312938755 @default.
- W4312121097 isParatext "false" @default.
- W4312121097 isRetracted "false" @default.
- W4312121097 workType "article" @default.