Matches in SemOpenAlex for { <https://semopenalex.org/work/W2524428287> ?p ?o ?g. }
- W2524428287 endingPage "6898" @default.
- W2524428287 startingPage "6869" @default.
- W2524428287 abstract "We introduce a method to train Quantized Neural Networks (QNNs) -- neural networks with extremely low precision (e.g., 1-bit) weights and activations, at run-time. At traintime the quantized weights and activations are used for computing the parameter gradients. During the forward pass, QNNs drastically reduce memory size and accesses, and replace most arithmetic operations with bit-wise operations. As a result, power consumption is expected to be drastically reduced. We trained QNNs over the MNIST, CIFAR-10, SVHN and ImageNet datasets. The resulting QNNs achieve prediction accuracy comparable to their 32-bit counterparts. For example, our quantized version of AlexNet with 1-bit weights and 2-bit activations achieves 51% top-1 accuracy. Moreover, we quantize the parameter gradients to 6-bits as well which enables gradients computation using only bit-wise operation. Quantized recurrent neural networks were tested over the Penn Treebank dataset, and achieved comparable accuracy as their 32-bit counterparts using only 4-bits. Last but not least, we programmed a binary matrix multiplication GPU kernel with which it is possible to run our MNIST QNN 7 times faster than with an unoptimized GPU kernel, without suffering any loss in classification accuracy. The QNN code is available online." @default.
- W2524428287 created "2016-10-07" @default.
- W2524428287 creator A5025123281 @default.
- W2524428287 creator A5025669300 @default.
- W2524428287 creator A5039449234 @default.
- W2524428287 creator A5086198262 @default.
- W2524428287 creator A5087518651 @default.
- W2524428287 date "2017-01-01" @default.
- W2524428287 modified "2023-10-06" @default.
- W2524428287 title "Quantized neural networks: training neural networks with low precision weights and activations" @default.
- W2524428287 cites W1533861849 @default.
- W2524428287 cites W1546411676 @default.
- W2524428287 cites W1583776211 @default.
- W2524428287 cites W1598866093 @default.
- W2524428287 cites W1632114991 @default.
- W2524428287 cites W1724438581 @default.
- W2524428287 cites W1825672851 @default.
- W2524428287 cites W1836465849 @default.
- W2524428287 cites W1841592590 @default.
- W2524428287 cites W1845051632 @default.
- W2524428287 cites W1861262881 @default.
- W2524428287 cites W1902041153 @default.
- W2524428287 cites W1902934009 @default.
- W2524428287 cites W1999085092 @default.
- W2524428287 cites W1999965501 @default.
- W2524428287 cites W2013305145 @default.
- W2524428287 cites W2064675550 @default.
- W2524428287 cites W2095705004 @default.
- W2524428287 cites W2108677974 @default.
- W2524428287 cites W2112739286 @default.
- W2524428287 cites W2112796928 @default.
- W2524428287 cites W2125203716 @default.
- W2524428287 cites W2152839228 @default.
- W2524428287 cites W2160321753 @default.
- W2524428287 cites W2161758346 @default.
- W2524428287 cites W2162390675 @default.
- W2524428287 cites W2168894214 @default.
- W2524428287 cites W2183631084 @default.
- W2524428287 cites W2184045248 @default.
- W2524428287 cites W2211979669 @default.
- W2524428287 cites W2251682575 @default.
- W2524428287 cites W2257979135 @default.
- W2524428287 cites W2266701264 @default.
- W2524428287 cites W22861983 @default.
- W2524428287 cites W2294059674 @default.
- W2524428287 cites W2337344472 @default.
- W2524428287 cites W2418046303 @default.
- W2524428287 cites W2431931973 @default.
- W2524428287 cites W2469490737 @default.
- W2524428287 cites W2512629640 @default.
- W2524428287 cites W2618530766 @default.
- W2524428287 cites W2898422183 @default.
- W2524428287 cites W2949888546 @default.
- W2524428287 cites W2951978180 @default.
- W2524428287 cites W2952432176 @default.
- W2524428287 cites W2962835968 @default.
- W2524428287 cites W2963919294 @default.
- W2524428287 cites W2964121744 @default.
- W2524428287 cites W2964299589 @default.
- W2524428287 cites W2964308564 @default.
- W2524428287 cites W4919037 @default.
- W2524428287 cites W587794757 @default.
- W2524428287 cites W753012316 @default.
- W2524428287 hasPublicationYear "2017" @default.
- W2524428287 type Work @default.
- W2524428287 sameAs 2524428287 @default.
- W2524428287 citedByCount "403" @default.
- W2524428287 countsByYear W25244282872016 @default.
- W2524428287 countsByYear W25244282872017 @default.
- W2524428287 countsByYear W25244282872018 @default.
- W2524428287 countsByYear W25244282872019 @default.
- W2524428287 countsByYear W25244282872020 @default.
- W2524428287 countsByYear W25244282872021 @default.
- W2524428287 countsByYear W25244282872022 @default.
- W2524428287 countsByYear W25244282872023 @default.
- W2524428287 crossrefType "journal-article" @default.
- W2524428287 hasAuthorship W2524428287A5025123281 @default.
- W2524428287 hasAuthorship W2524428287A5025669300 @default.
- W2524428287 hasAuthorship W2524428287A5039449234 @default.
- W2524428287 hasAuthorship W2524428287A5086198262 @default.
- W2524428287 hasAuthorship W2524428287A5087518651 @default.
- W2524428287 hasConcept C11413529 @default.
- W2524428287 hasConcept C114614502 @default.
- W2524428287 hasConcept C154945302 @default.
- W2524428287 hasConcept C177264268 @default.
- W2524428287 hasConcept C190502265 @default.
- W2524428287 hasConcept C19768560 @default.
- W2524428287 hasConcept C199360897 @default.
- W2524428287 hasConcept C206134035 @default.
- W2524428287 hasConcept C2776760102 @default.
- W2524428287 hasConcept C2780595030 @default.
- W2524428287 hasConcept C33923547 @default.
- W2524428287 hasConcept C41008148 @default.
- W2524428287 hasConcept C45374587 @default.
- W2524428287 hasConcept C48372109 @default.
- W2524428287 hasConcept C50644808 @default.
- W2524428287 hasConcept C74193536 @default.
- W2524428287 hasConcept C94375191 @default.