File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Bayesian Neural Network Realization by Exploiting Inherent Stochastic Characteristics of Analog RRAM

TitleBayesian Neural Network Realization by Exploiting Inherent Stochastic Characteristics of Analog RRAM
Authors
Issue Date2019
Citation
Technical Digest - International Electron Devices Meeting, IEDM, 2019, v. 2019-December, article no. 8993616 How to Cite?
AbstractFor the first time, this paper develops a novel stochastic computing method by utilizing the inherent random noises of analog RRAM. With the designed analog switching characteristics, the RRAM device can realize the function of sampling from a tunable probabilistic distribution. A Bayesian neural network (BayNN), whose weights are represented by probability distributions, is experimentally demonstrated on the fabricated 160K RRAM array. The measured result achieves 97% accuracy for image classification on MNIST dataset. Moreover, the RRAM based BayNN shows anti-attack capability with inherent device stochastic behavior to detect "adversarial" images, which are generated by adding noises to the original MNIST images and can fool the conventional deep neural networks. This is the first demonstration work for the widely-used BayNN algorithms with emerging devices.
Persistent Identifierhttp://hdl.handle.net/10722/334645
ISSN
2023 SCImago Journal Rankings: 1.047
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorLin, Yudeng-
dc.contributor.authorHu, Xiaobo Sharon-
dc.contributor.authorQian, He-
dc.contributor.authorWu, Huaqiang-
dc.contributor.authorZhang, Qingtian-
dc.contributor.authorTang, Jianshi-
dc.contributor.authorGao, Bin-
dc.contributor.authorLi, Chongxuan-
dc.contributor.authorYao, Peng-
dc.contributor.authorLiu, Zhengwu-
dc.contributor.authorZhu, Jun-
dc.contributor.authorLu, Jiwu-
dc.date.accessioned2023-10-20T06:49:37Z-
dc.date.available2023-10-20T06:49:37Z-
dc.date.issued2019-
dc.identifier.citationTechnical Digest - International Electron Devices Meeting, IEDM, 2019, v. 2019-December, article no. 8993616-
dc.identifier.issn0163-1918-
dc.identifier.urihttp://hdl.handle.net/10722/334645-
dc.description.abstractFor the first time, this paper develops a novel stochastic computing method by utilizing the inherent random noises of analog RRAM. With the designed analog switching characteristics, the RRAM device can realize the function of sampling from a tunable probabilistic distribution. A Bayesian neural network (BayNN), whose weights are represented by probability distributions, is experimentally demonstrated on the fabricated 160K RRAM array. The measured result achieves 97% accuracy for image classification on MNIST dataset. Moreover, the RRAM based BayNN shows anti-attack capability with inherent device stochastic behavior to detect "adversarial" images, which are generated by adding noises to the original MNIST images and can fool the conventional deep neural networks. This is the first demonstration work for the widely-used BayNN algorithms with emerging devices.-
dc.languageeng-
dc.relation.ispartofTechnical Digest - International Electron Devices Meeting, IEDM-
dc.titleBayesian Neural Network Realization by Exploiting Inherent Stochastic Characteristics of Analog RRAM-
dc.typeConference_Paper-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1109/IEDM19573.2019.8993616-
dc.identifier.scopuseid_2-s2.0-85081046185-
dc.identifier.volume2019-December-
dc.identifier.spagearticle no. 8993616-
dc.identifier.epagearticle no. 8993616-
dc.identifier.isiWOS:000553550000182-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats