Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204278478> ?p ?o ?g. }
Showing items 1 to 90 of
90
with 100 items per page.
- W3204278478 abstract "Low-precision computation, which has been widely supported in contemporary hardware, is considered as one of the most effective methods to accelerate convolutional neural networks. However, low-precision computation is not widely used to speed up Winograd, an algorithm for fast convolution computation, due to the numerical error introduced by combining Winograd transformation and quantization. In this paper, we propose a low-precision Winograd convolution approach, LoWino, based on post-training quantization, which employs a linear quantization method in the Winograd domain to reduce the precision loss caused by transformations. Moreover, we present an efficient implementation that integrates well-designed optimization techniques, thereby adequately exploiting the capability of low-precision computation on modern CPUs. We evaluate our approach on Intel Xeon Scalable Processors by leveraging representative convolutional layers in prevailing deep neural networks. Experimental results show that LoWino achieves up to 2.04 × speedup over state-of-the-art implementations in the vendor library while maintaining the accuracy at a reasonable level." @default.
- W3204278478 created "2021-10-11" @default.
- W3204278478 creator A5004017901 @default.
- W3204278478 creator A5023640358 @default.
- W3204278478 creator A5035186499 @default.
- W3204278478 creator A5053070701 @default.
- W3204278478 date "2021-08-09" @default.
- W3204278478 modified "2023-09-24" @default.
- W3204278478 title "LoWino: Towards Efficient Low-Precision Winograd Convolutions on Modern CPUs" @default.
- W3204278478 cites W1487564550 @default.
- W3204278478 cites W1901129140 @default.
- W3204278478 cites W2065710348 @default.
- W3204278478 cites W2097117768 @default.
- W3204278478 cites W2108598243 @default.
- W3204278478 cites W2172654076 @default.
- W3204278478 cites W2194775991 @default.
- W3204278478 cites W2540279855 @default.
- W3204278478 cites W2618939455 @default.
- W3204278478 cites W2786374423 @default.
- W3204278478 cites W2787884921 @default.
- W3204278478 cites W2950649068 @default.
- W3204278478 cites W2963122961 @default.
- W3204278478 cites W2966126880 @default.
- W3204278478 cites W2997109118 @default.
- W3204278478 cites W2998218113 @default.
- W3204278478 cites W2998957070 @default.
- W3204278478 cites W3008378296 @default.
- W3204278478 cites W3015869703 @default.
- W3204278478 cites W3018661492 @default.
- W3204278478 cites W3098382995 @default.
- W3204278478 doi "https://doi.org/10.1145/3472456.3472464" @default.
- W3204278478 hasPublicationYear "2021" @default.
- W3204278478 type Work @default.
- W3204278478 sameAs 3204278478 @default.
- W3204278478 citedByCount "4" @default.
- W3204278478 countsByYear W32042784782022 @default.
- W3204278478 countsByYear W32042784782023 @default.
- W3204278478 crossrefType "proceedings-article" @default.
- W3204278478 hasAuthorship W3204278478A5004017901 @default.
- W3204278478 hasAuthorship W3204278478A5023640358 @default.
- W3204278478 hasAuthorship W3204278478A5035186499 @default.
- W3204278478 hasAuthorship W3204278478A5053070701 @default.
- W3204278478 hasBestOaLocation W32042784781 @default.
- W3204278478 hasConcept C113775141 @default.
- W3204278478 hasConcept C11413529 @default.
- W3204278478 hasConcept C145108525 @default.
- W3204278478 hasConcept C154945302 @default.
- W3204278478 hasConcept C173608175 @default.
- W3204278478 hasConcept C28855332 @default.
- W3204278478 hasConcept C41008148 @default.
- W3204278478 hasConcept C45347329 @default.
- W3204278478 hasConcept C45374587 @default.
- W3204278478 hasConcept C48044578 @default.
- W3204278478 hasConcept C50644808 @default.
- W3204278478 hasConcept C68339613 @default.
- W3204278478 hasConcept C77088390 @default.
- W3204278478 hasConcept C81363708 @default.
- W3204278478 hasConcept C96972482 @default.
- W3204278478 hasConceptScore W3204278478C113775141 @default.
- W3204278478 hasConceptScore W3204278478C11413529 @default.
- W3204278478 hasConceptScore W3204278478C145108525 @default.
- W3204278478 hasConceptScore W3204278478C154945302 @default.
- W3204278478 hasConceptScore W3204278478C173608175 @default.
- W3204278478 hasConceptScore W3204278478C28855332 @default.
- W3204278478 hasConceptScore W3204278478C41008148 @default.
- W3204278478 hasConceptScore W3204278478C45347329 @default.
- W3204278478 hasConceptScore W3204278478C45374587 @default.
- W3204278478 hasConceptScore W3204278478C48044578 @default.
- W3204278478 hasConceptScore W3204278478C50644808 @default.
- W3204278478 hasConceptScore W3204278478C68339613 @default.
- W3204278478 hasConceptScore W3204278478C77088390 @default.
- W3204278478 hasConceptScore W3204278478C81363708 @default.
- W3204278478 hasConceptScore W3204278478C96972482 @default.
- W3204278478 hasLocation W32042784781 @default.
- W3204278478 hasOpenAccess W3204278478 @default.
- W3204278478 hasPrimaryLocation W32042784781 @default.
- W3204278478 hasRelatedWork W123738650 @default.
- W3204278478 hasRelatedWork W1587475692 @default.
- W3204278478 hasRelatedWork W2028244810 @default.
- W3204278478 hasRelatedWork W2074129693 @default.
- W3204278478 hasRelatedWork W2385361820 @default.
- W3204278478 hasRelatedWork W2392910503 @default.
- W3204278478 hasRelatedWork W2965967938 @default.
- W3204278478 hasRelatedWork W3045822024 @default.
- W3204278478 hasRelatedWork W4251418457 @default.
- W3204278478 hasRelatedWork W45427617 @default.
- W3204278478 isParatext "false" @default.
- W3204278478 isRetracted "false" @default.
- W3204278478 magId "3204278478" @default.
- W3204278478 workType "article" @default.