FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4285798357> ?p ?o ?g. }

• W4285798357 endingPage "24" @default.
• W4285798357 startingPage "1" @default.
• W4285798357 abstract "Deep Neural Networks (DNNs) have achieved remarkable success in various Artificial Intelligence applications. Quantization is a critical step in DNNs compression and acceleration for deployment. To further boost DNN execution efficiency, many works explore to leverage the input-dependent redundancy with dynamic quantization for different regions. However, the sensitive regions in the feature map are irregularly distributed, which restricts the real speed up for existing accelerators. To this end, we propose an algorithm-architecture co-design, named Structured Dynamic Precision (SDP). Specifically, we propose a quantization scheme in which the high-order bit part and the low-order bit part of data can be masked independently. And a fixed number of term parts are dynamically selected for computation based on the importance of each term in the group. We also present a hardware design to enable the algorithm efficiently with small overheads, whose inference time mainly scales with the precision proportionally. Evaluation experiments on extensive networks demonstrate that compared to the state-of-the-art dynamic quantization accelerator DRQ, our SDP can achieve 29% performance gain and 51% energy reduction for the same level of model accuracy." @default.
• W4285798357 created "2022-07-19" @default.
• W4285798357 creator A5004005307 @default.
• W4285798357 creator A5008244796 @default.
• W4285798357 creator A5017537432 @default.
• W4285798357 creator A5021783986 @default.
• W4285798357 creator A5023776065 @default.
• W4285798357 creator A5042525419 @default.
• W4285798357 creator A5063591197 @default.
• W4285798357 creator A5066733526 @default.
• W4285798357 creator A5072312399 @default.
• W4285798357 creator A5073803879 @default.
• W4285798357 date "2023-01-20" @default.
• W4285798357 modified "2023-09-26" @default.
• W4285798357 title "Structured Dynamic Precision for Deep Neural Networks Quantization" @default.
• W4285798357 cites W1885185971 @default.
• W4285798357 cites W2048266589 @default.
• W4285798357 cites W2102605133 @default.
• W4285798357 cites W2108598243 @default.
• W4285798357 cites W2162639668 @default.
• W4285798357 cites W2194775991 @default.
• W4285798357 cites W2442974303 @default.
• W4285798357 cites W2541839172 @default.
• W4285798357 cites W2606722458 @default.
• W4285798357 cites W2618530766 @default.
• W4285798357 cites W2884071170 @default.
• W4285798357 cites W2905184098 @default.
• W4285798357 cites W2949870694 @default.
• W4285798357 cites W2963163009 @default.
• W4285798357 cites W2963367920 @default.
• W4285798357 cites W2979515806 @default.
• W4285798357 cites W3012561096 @default.
• W4285798357 doi "https://doi.org/10.1145/3549535" @default.
• W4285798357 hasPublicationYear "2023" @default.
• W4285798357 type Work @default.
• W4285798357 citedByCount "0" @default.
• W4285798357 crossrefType "journal-article" @default.
• W4285798357 hasAuthorship W4285798357A5004005307 @default.
• W4285798357 hasAuthorship W4285798357A5008244796 @default.
• W4285798357 hasAuthorship W4285798357A5017537432 @default.
• W4285798357 hasAuthorship W4285798357A5021783986 @default.
• W4285798357 hasAuthorship W4285798357A5023776065 @default.
• W4285798357 hasAuthorship W4285798357A5042525419 @default.
• W4285798357 hasAuthorship W4285798357A5063591197 @default.
• W4285798357 hasAuthorship W4285798357A5066733526 @default.
• W4285798357 hasAuthorship W4285798357A5072312399 @default.
• W4285798357 hasAuthorship W4285798357A5073803879 @default.
• W4285798357 hasConcept C111919701 @default.
• W4285798357 hasConcept C113775141 @default.
• W4285798357 hasConcept C11413529 @default.
• W4285798357 hasConcept C119599485 @default.
• W4285798357 hasConcept C127413603 @default.
• W4285798357 hasConcept C13164978 @default.
• W4285798357 hasConcept C152124472 @default.
• W4285798357 hasConcept C153083717 @default.
• W4285798357 hasConcept C154945302 @default.
• W4285798357 hasConcept C2742236 @default.
• W4285798357 hasConcept C2776214188 @default.
• W4285798357 hasConcept C28855332 @default.
• W4285798357 hasConcept C2984842247 @default.
• W4285798357 hasConcept C41008148 @default.
• W4285798357 hasConcept C42935608 @default.
• W4285798357 hasConcept C45374587 @default.
• W4285798357 hasConcept C50644808 @default.
• W4285798357 hasConcept C9390403 @default.
• W4285798357 hasConceptScore W4285798357C111919701 @default.
• W4285798357 hasConceptScore W4285798357C113775141 @default.
• W4285798357 hasConceptScore W4285798357C11413529 @default.
• W4285798357 hasConceptScore W4285798357C119599485 @default.
• W4285798357 hasConceptScore W4285798357C127413603 @default.
• W4285798357 hasConceptScore W4285798357C13164978 @default.
• W4285798357 hasConceptScore W4285798357C152124472 @default.
• W4285798357 hasConceptScore W4285798357C153083717 @default.
• W4285798357 hasConceptScore W4285798357C154945302 @default.
• W4285798357 hasConceptScore W4285798357C2742236 @default.
• W4285798357 hasConceptScore W4285798357C2776214188 @default.
• W4285798357 hasConceptScore W4285798357C28855332 @default.
• W4285798357 hasConceptScore W4285798357C2984842247 @default.
• W4285798357 hasConceptScore W4285798357C41008148 @default.
• W4285798357 hasConceptScore W4285798357C42935608 @default.
• W4285798357 hasConceptScore W4285798357C45374587 @default.
• W4285798357 hasConceptScore W4285798357C50644808 @default.
• W4285798357 hasConceptScore W4285798357C9390403 @default.
• W4285798357 hasFunder F4320335777 @default.
• W4285798357 hasIssue "1" @default.
• W4285798357 hasLocation W42857983571 @default.
• W4285798357 hasOpenAccess W4285798357 @default.
• W4285798357 hasPrimaryLocation W42857983571 @default.
• W4285798357 hasRelatedWork W2777406049 @default.
• W4285798357 hasRelatedWork W2799290373 @default.
• W4285798357 hasRelatedWork W2803935332 @default.
• W4285798357 hasRelatedWork W2898755250 @default.
• W4285798357 hasRelatedWork W3037896803 @default.
• W4285798357 hasRelatedWork W3099092507 @default.
• W4285798357 hasRelatedWork W3196579076 @default.
• W4285798357 hasRelatedWork W3217204427 @default.
• W4285798357 hasRelatedWork W4310823283 @default.
• W4285798357 hasRelatedWork W4319430125 @default.

• next