Soft-Sign Stochastic Gradient Descent Algorithm for Wireless Federated Learning
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Lee, Seunghoon | - |
dc.contributor.author | Park, Chanho | - |
dc.contributor.author | Hong, Songnam | - |
dc.contributor.author | Eldar, Yonina C. | - |
dc.contributor.author | Lee, Namyoon | - |
dc.date.accessioned | 2022-07-06T11:33:39Z | - |
dc.date.available | 2022-07-06T11:33:39Z | - |
dc.date.created | 2022-01-26 | - |
dc.date.issued | 2021-11 | - |
dc.identifier.issn | 0000-0000 | - |
dc.identifier.uri | https://scholarworks.bwise.kr/hanyang/handle/2021.sw.hanyang/140382 | - |
dc.description.abstract | Federated learning over wireless networks requires aggregating locally computed gradients at a server where the mobile devices send statistically distinct gradient information over heterogenous communication links. This paper proposes a Bayesian approach for wireless federated learning referred to as soft-sign stochastic gradient descent (soft-signSGD). The idea of soft-signSGD is to aggregate the one-bit quantized local gradients at the server by jointly exploiting i) the prior distributions of the local gradients, ii) the gradient quantizer function, and iii) channel distributions. This aggregation method is optimal in the sense of minimizing the mean-squared error (MSE) under a simplified Gaussian prior assumption on the local gradient. From simulations, we demonstrate that soft-signSGD considerably outperforms the conventional sign stochastic gradient descent algorithm when training and testing neural networks using the MNIST dataset and the CIFAR-10 dataset over heterogeneous wireless networks. | - |
dc.language | 영어 | - |
dc.language.iso | en | - |
dc.publisher | Institute of Electrical and Electronics Engineers Inc. | - |
dc.title | Soft-Sign Stochastic Gradient Descent Algorithm for Wireless Federated Learning | - |
dc.type | Article | - |
dc.contributor.affiliatedAuthor | Hong, Songnam | - |
dc.identifier.doi | 10.1109/SPAWC51858.2021.9593212 | - |
dc.identifier.scopusid | 2-s2.0-85122822877 | - |
dc.identifier.bibliographicCitation | IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC, v.2021, no.September, pp.241 - 245 | - |
dc.relation.isPartOf | IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC | - |
dc.citation.title | IEEE Workshop on Signal Processing Advances in Wireless Communications, SPAWC | - |
dc.citation.volume | 2021 | - |
dc.citation.number | September | - |
dc.citation.startPage | 241 | - |
dc.citation.endPage | 245 | - |
dc.type.rims | ART | - |
dc.type.docType | Conference Paper | - |
dc.description.journalClass | 1 | - |
dc.description.isOpenAccess | N | - |
dc.description.journalRegisteredClass | scopus | - |
dc.subject.keywordPlus | Bayesian networks | - |
dc.subject.keywordPlus | Gradient methods | - |
dc.subject.keywordPlus | Heterogeneous networks | - |
dc.subject.keywordPlus | Statistical tests | - |
dc.subject.keywordPlus | Stochastic systems | - |
dc.subject.keywordPlus | Wireless networks | - |
dc.subject.keywordPlus | Aggregation methods | - |
dc.subject.keywordPlus | Bayesian approaches | - |
dc.subject.keywordPlus | Channel distributions | - |
dc.subject.keywordPlus | Gradient informations | - |
dc.subject.keywordPlus | Local gradients | - |
dc.subject.keywordPlus | Mean squared error | - |
dc.subject.keywordPlus | Prior distribution | - |
dc.subject.keywordPlus | Quantizers | - |
dc.subject.keywordPlus | Stochastic gradient descent | - |
dc.subject.keywordPlus | Stochastic gradient descent algorithm | - |
dc.subject.keywordPlus | Mean square error | - |
dc.identifier.url | https://ieeexplore.ieee.org/document/9593212 | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
222, Wangsimni-ro, Seongdong-gu, Seoul, 04763, Korea+82-2-2220-1365
COPYRIGHT © 2021 HANYANG UNIVERSITY.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.