Matches in SemOpenAlex for { <https://semopenalex.org/work/W4229445200> ?p ?o ?g. }
Showing items 1 to 99 of
99
with 100 items per page.
- W4229445200 endingPage "108962" @default.
- W4229445200 startingPage "108962" @default.
- W4229445200 abstract "With the increasing requirements for the combination of software and hardware, network compression and hardware deployment have become hot research topics. In network compression, binary neural networks (BNNs) are widely applied in artificial intelligence chips because of memory saving, high computational efficiency, and hardware friendliness. However, there is a performance gap between BNNs and full-precision neural networks (FNNs). This paper proposes a BNN training framework called KDG-BNN, consisting of three modules: a full-precision network, a 1-bit binary network, and a discriminator. The full-precision network guides the 1-bit binary network to train through distillation loss in this framework. Meanwhile, the 1-bit binary network acts as a generator and conducts adversarial training with the discriminator. By simultaneously optimizing the adversarial loss and distillation loss, the 1-bit binary network can learn the feature distribution of the full-precision network more accurately. Then, the generative adversarial network (GAN) is replaced by Wasserstein GAN with gradient penalty (WGAN-GP) to deal with gradient disappearance, and KDG-BNN is developed into KDWG-BNN. Experiments show that AdamBNN trained with KDWG-BNN can achieve 85.89% and 70.7% accuracy on CIFAR-10 and ImageNet, respectively, exceeding 0.76% on CIFAR-10 and 0.2% on ImageNet. The memristor has many features for hardware deployment, such as memory functions, continuous input and output, nanoscale size, etc., making it an ideal device for deploying neural networks. Therefore, this paper further proposes a memristor-based KDG-BNN implementation scheme by levering the merits of memristors and the lightweight BNNs in the hope of realizing and promoting end-side intelligent applications." @default.
- W4229445200 created "2022-05-11" @default.
- W4229445200 creator A5032754937 @default.
- W4229445200 creator A5035048973 @default.
- W4229445200 creator A5059395980 @default.
- W4229445200 creator A5087454088 @default.
- W4229445200 date "2022-08-01" @default.
- W4229445200 modified "2023-10-17" @default.
- W4229445200 title "Memristive KDG-BNN: Memristive binary neural networks trained via knowledge distillation and generative adversarial networks" @default.
- W4229445200 cites W1536680647 @default.
- W4229445200 cites W2008901850 @default.
- W4229445200 cites W2016922062 @default.
- W4229445200 cites W2112181056 @default.
- W4229445200 cites W2117539524 @default.
- W4229445200 cites W2162651880 @default.
- W4229445200 cites W2194775991 @default.
- W4229445200 cites W2379133119 @default.
- W4229445200 cites W2887447938 @default.
- W4229445200 cites W2894994475 @default.
- W4229445200 cites W2953497520 @default.
- W4229445200 cites W2963125010 @default.
- W4229445200 cites W2994356318 @default.
- W4229445200 cites W3008515144 @default.
- W4229445200 cites W3010557233 @default.
- W4229445200 cites W3034297393 @default.
- W4229445200 cites W3034681682 @default.
- W4229445200 cites W3037030727 @default.
- W4229445200 cites W3083318754 @default.
- W4229445200 cites W3121050449 @default.
- W4229445200 cites W3122126208 @default.
- W4229445200 cites W3132454634 @default.
- W4229445200 cites W3144024160 @default.
- W4229445200 cites W3159420753 @default.
- W4229445200 cites W3167583064 @default.
- W4229445200 cites W3174261480 @default.
- W4229445200 cites W3176211720 @default.
- W4229445200 cites W3181161645 @default.
- W4229445200 cites W3199986538 @default.
- W4229445200 cites W3203249428 @default.
- W4229445200 cites W3208963443 @default.
- W4229445200 doi "https://doi.org/10.1016/j.knosys.2022.108962" @default.
- W4229445200 hasPublicationYear "2022" @default.
- W4229445200 type Work @default.
- W4229445200 citedByCount "5" @default.
- W4229445200 countsByYear W42294452002022 @default.
- W4229445200 countsByYear W42294452002023 @default.
- W4229445200 crossrefType "journal-article" @default.
- W4229445200 hasAuthorship W4229445200A5032754937 @default.
- W4229445200 hasAuthorship W4229445200A5035048973 @default.
- W4229445200 hasAuthorship W4229445200A5059395980 @default.
- W4229445200 hasAuthorship W4229445200A5087454088 @default.
- W4229445200 hasConcept C113775141 @default.
- W4229445200 hasConcept C127413603 @default.
- W4229445200 hasConcept C150072547 @default.
- W4229445200 hasConcept C154945302 @default.
- W4229445200 hasConcept C24326235 @default.
- W4229445200 hasConcept C2779803651 @default.
- W4229445200 hasConcept C33923547 @default.
- W4229445200 hasConcept C41008148 @default.
- W4229445200 hasConcept C48372109 @default.
- W4229445200 hasConcept C50644808 @default.
- W4229445200 hasConcept C76155785 @default.
- W4229445200 hasConcept C94375191 @default.
- W4229445200 hasConcept C94915269 @default.
- W4229445200 hasConceptScore W4229445200C113775141 @default.
- W4229445200 hasConceptScore W4229445200C127413603 @default.
- W4229445200 hasConceptScore W4229445200C150072547 @default.
- W4229445200 hasConceptScore W4229445200C154945302 @default.
- W4229445200 hasConceptScore W4229445200C24326235 @default.
- W4229445200 hasConceptScore W4229445200C2779803651 @default.
- W4229445200 hasConceptScore W4229445200C33923547 @default.
- W4229445200 hasConceptScore W4229445200C41008148 @default.
- W4229445200 hasConceptScore W4229445200C48372109 @default.
- W4229445200 hasConceptScore W4229445200C50644808 @default.
- W4229445200 hasConceptScore W4229445200C76155785 @default.
- W4229445200 hasConceptScore W4229445200C94375191 @default.
- W4229445200 hasConceptScore W4229445200C94915269 @default.
- W4229445200 hasFunder F4320321001 @default.
- W4229445200 hasFunder F4320323172 @default.
- W4229445200 hasFunder F4320335777 @default.
- W4229445200 hasLocation W42294452001 @default.
- W4229445200 hasOpenAccess W4229445200 @default.
- W4229445200 hasPrimaryLocation W42294452001 @default.
- W4229445200 hasRelatedWork W1892057141 @default.
- W4229445200 hasRelatedWork W2019421138 @default.
- W4229445200 hasRelatedWork W2119690829 @default.
- W4229445200 hasRelatedWork W2132696896 @default.
- W4229445200 hasRelatedWork W2386387936 @default.
- W4229445200 hasRelatedWork W2767624311 @default.
- W4229445200 hasRelatedWork W2972430781 @default.
- W4229445200 hasRelatedWork W3215828240 @default.
- W4229445200 hasRelatedWork W4280544492 @default.
- W4229445200 hasRelatedWork W4312963184 @default.
- W4229445200 hasVolume "249" @default.
- W4229445200 isParatext "false" @default.
- W4229445200 isRetracted "false" @default.
- W4229445200 workType "article" @default.