Matches in SemOpenAlex for { <https://semopenalex.org/work/W3030474951> ?p ?o ?g. }
Showing items 1 to 100 of
100
with 100 items per page.
- W3030474951 abstract "The inherent heavy computation of deep neural networks prevents their widespread applications. A widely used method for accelerating model inference is quantization, by replacing the input operands of a network using fixed-point values. Then the majority of computation costs focus on the integer matrix multiplication accumulation. In fact, high-bit accumulator leads to partially wasted computation and low-bit one typically suffers from numerical overflow. To address this problem, we propose an overflow aware quantization method by designing trainable adaptive fixed-point representation, to optimize the number of bits for each input tensor while prohibiting numeric overflow during the computation. With the proposed method, we are able to fully utilize the computing power to minimize the quantization loss and obtain optimized inference performance. To verify the effectiveness of our method, we conduct image classification, object detection, and semantic segmentation tasks on ImageNet, Pascal VOC, and COCO datasets, respectively. Experimental results demonstrate that the proposed method can achieve comparable performance with state-of-the-art quantization methods while accelerating the inference process by about 2 times." @default.
- W3030474951 created "2020-06-05" @default.
- W3030474951 creator A5026519540 @default.
- W3030474951 creator A5032618881 @default.
- W3030474951 creator A5063997218 @default.
- W3030474951 creator A5066748214 @default.
- W3030474951 creator A5077183319 @default.
- W3030474951 creator A5077972623 @default.
- W3030474951 creator A5082924617 @default.
- W3030474951 creator A5090651667 @default.
- W3030474951 date "2020-05-27" @default.
- W3030474951 modified "2023-09-26" @default.
- W3030474951 title "Accelerating Neural Network Inference by Overflow Aware Quantization" @default.
- W3030474951 cites W2163605009 @default.
- W3030474951 cites W2194775991 @default.
- W3030474951 cites W2612445135 @default.
- W3030474951 cites W2613718673 @default.
- W3030474951 cites W2786771851 @default.
- W3030474951 cites W2793950911 @default.
- W3030474951 cites W2809624076 @default.
- W3030474951 cites W2894946950 @default.
- W3030474951 cites W2903260438 @default.
- W3030474951 cites W2944508492 @default.
- W3030474951 cites W2946948417 @default.
- W3030474951 cites W2960643108 @default.
- W3030474951 cites W2963122961 @default.
- W3030474951 cites W2963163009 @default.
- W3030474951 cites W2964259004 @default.
- W3030474951 cites W2964309882 @default.
- W3030474951 cites W2964861433 @default.
- W3030474951 cites W2966153044 @default.
- W3030474951 cites W2972918064 @default.
- W3030474951 cites W2982479999 @default.
- W3030474951 cites W3004061291 @default.
- W3030474951 cites W3038006402 @default.
- W3030474951 cites W3106250896 @default.
- W3030474951 hasPublicationYear "2020" @default.
- W3030474951 type Work @default.
- W3030474951 sameAs 3030474951 @default.
- W3030474951 citedByCount "0" @default.
- W3030474951 crossrefType "posted-content" @default.
- W3030474951 hasAuthorship W3030474951A5026519540 @default.
- W3030474951 hasAuthorship W3030474951A5032618881 @default.
- W3030474951 hasAuthorship W3030474951A5063997218 @default.
- W3030474951 hasAuthorship W3030474951A5066748214 @default.
- W3030474951 hasAuthorship W3030474951A5077183319 @default.
- W3030474951 hasAuthorship W3030474951A5077972623 @default.
- W3030474951 hasAuthorship W3030474951A5082924617 @default.
- W3030474951 hasAuthorship W3030474951A5090651667 @default.
- W3030474951 hasConcept C11413529 @default.
- W3030474951 hasConcept C154945302 @default.
- W3030474951 hasConcept C173608175 @default.
- W3030474951 hasConcept C199360897 @default.
- W3030474951 hasConcept C2776214188 @default.
- W3030474951 hasConcept C28855332 @default.
- W3030474951 hasConcept C41008148 @default.
- W3030474951 hasConcept C45374587 @default.
- W3030474951 hasConcept C50644808 @default.
- W3030474951 hasConcept C68339613 @default.
- W3030474951 hasConcept C75608658 @default.
- W3030474951 hasConcept C84211073 @default.
- W3030474951 hasConceptScore W3030474951C11413529 @default.
- W3030474951 hasConceptScore W3030474951C154945302 @default.
- W3030474951 hasConceptScore W3030474951C173608175 @default.
- W3030474951 hasConceptScore W3030474951C199360897 @default.
- W3030474951 hasConceptScore W3030474951C2776214188 @default.
- W3030474951 hasConceptScore W3030474951C28855332 @default.
- W3030474951 hasConceptScore W3030474951C41008148 @default.
- W3030474951 hasConceptScore W3030474951C45374587 @default.
- W3030474951 hasConceptScore W3030474951C50644808 @default.
- W3030474951 hasConceptScore W3030474951C68339613 @default.
- W3030474951 hasConceptScore W3030474951C75608658 @default.
- W3030474951 hasConceptScore W3030474951C84211073 @default.
- W3030474951 hasLocation W30304749511 @default.
- W3030474951 hasOpenAccess W3030474951 @default.
- W3030474951 hasPrimaryLocation W30304749511 @default.
- W3030474951 hasRelatedWork W2497335382 @default.
- W3030474951 hasRelatedWork W2777406049 @default.
- W3030474951 hasRelatedWork W2806978588 @default.
- W3030474951 hasRelatedWork W2891335318 @default.
- W3030474951 hasRelatedWork W2964472257 @default.
- W3030474951 hasRelatedWork W2970472742 @default.
- W3030474951 hasRelatedWork W2971667683 @default.
- W3030474951 hasRelatedWork W3001269559 @default.
- W3030474951 hasRelatedWork W3008684984 @default.
- W3030474951 hasRelatedWork W3044757432 @default.
- W3030474951 hasRelatedWork W3103209792 @default.
- W3030474951 hasRelatedWork W3113797829 @default.
- W3030474951 hasRelatedWork W3123666279 @default.
- W3030474951 hasRelatedWork W3135784973 @default.
- W3030474951 hasRelatedWork W3139173542 @default.
- W3030474951 hasRelatedWork W3143107579 @default.
- W3030474951 hasRelatedWork W3159072220 @default.
- W3030474951 hasRelatedWork W3172749677 @default.
- W3030474951 hasRelatedWork W3181793056 @default.
- W3030474951 hasRelatedWork W3214250042 @default.
- W3030474951 isParatext "false" @default.
- W3030474951 isRetracted "false" @default.
- W3030474951 magId "3030474951" @default.
- W3030474951 workType "article" @default.