Efficient low-rank federated learning based on singular value decomposition

Jungmin Kwon; Hyunggon Park

doi:10.1145/3492866.3561258

Efficient low-rank federated learning based on singular value decomposition

Jungmin Kwon, Hyunggon Park

Electronic and Electrical Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

In this paper, we propose a low-rank federated learning (FL) algorithm based on singular value decomposition (SVD). The SVD factorizes the global parameters that need to be exchanged between a global server and clients for distributed model training, significantly reducing the associated communication cost. Experiment results confirm that the number of transmissions is significantly reduced while maintaining the accuracy performance of the local model using the approximately recovered parameters.

Original language	English
Title of host publication	MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing
Publisher	Association for Computing Machinery
Pages	285-286
Number of pages	2
ISBN (Electronic)	9781450391658
DOIs	https://doi.org/10.1145/3492866.3561258
State	Published - 3 Oct 2022
Event	23rd ACM International Symposium on Mobile Ad Hoc Networking and Computing, MobiHoc 2022 - Seoul, Korea, Republic of Duration: 17 Oct 2022 → 20 Oct 2022

Publication series

Name	Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)

Conference

Conference	23rd ACM International Symposium on Mobile Ad Hoc Networking and Computing, MobiHoc 2022
Country/Territory	Korea, Republic of
City	Seoul
Period	17/10/22 → 20/10/22

Bibliographical note

Publisher Copyright:
© 2022 ACM.

Keywords

edge network
federated learning
low-rank matrix
singular value decomposition

Access to Document

10.1145/3492866.3561258

Cite this

Kwon, J., & Park, H. (2022). Efficient low-rank federated learning based on singular value decomposition. In MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing (pp. 285-286). (Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)). Association for Computing Machinery. https://doi.org/10.1145/3492866.3561258

Kwon, Jungmin ; Park, Hyunggon. / Efficient low-rank federated learning based on singular value decomposition. MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing. Association for Computing Machinery, 2022. pp. 285-286 (Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)).

@inproceedings{3aa60799fcd4460c811837b220b9268a,

title = "Efficient low-rank federated learning based on singular value decomposition",

abstract = "In this paper, we propose a low-rank federated learning (FL) algorithm based on singular value decomposition (SVD). The SVD factorizes the global parameters that need to be exchanged between a global server and clients for distributed model training, significantly reducing the associated communication cost. Experiment results confirm that the number of transmissions is significantly reduced while maintaining the accuracy performance of the local model using the approximately recovered parameters.",

keywords = "edge network, federated learning, low-rank matrix, singular value decomposition",

author = "Jungmin Kwon and Hyunggon Park",

note = "Publisher Copyright: {\textcopyright} 2022 ACM.; 23rd ACM International Symposium on Mobile Ad Hoc Networking and Computing, MobiHoc 2022 ; Conference date: 17-10-2022 Through 20-10-2022",

year = "2022",

month = oct,

day = "3",

doi = "10.1145/3492866.3561258",

language = "English",

series = "Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)",

publisher = "Association for Computing Machinery",

pages = "285--286",

booktitle = "MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing",

}

Kwon, J & Park, H 2022, Efficient low-rank federated learning based on singular value decomposition. in MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing. Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc), Association for Computing Machinery, pp. 285-286, 23rd ACM International Symposium on Mobile Ad Hoc Networking and Computing, MobiHoc 2022, Seoul, Korea, Republic of, 17/10/22. https://doi.org/10.1145/3492866.3561258

Efficient low-rank federated learning based on singular value decomposition. / Kwon, Jungmin; Park, Hyunggon.
MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing. Association for Computing Machinery, 2022. p. 285-286 (Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Efficient low-rank federated learning based on singular value decomposition

AU - Kwon, Jungmin

AU - Park, Hyunggon

PY - 2022/10/3

Y1 - 2022/10/3

N2 - In this paper, we propose a low-rank federated learning (FL) algorithm based on singular value decomposition (SVD). The SVD factorizes the global parameters that need to be exchanged between a global server and clients for distributed model training, significantly reducing the associated communication cost. Experiment results confirm that the number of transmissions is significantly reduced while maintaining the accuracy performance of the local model using the approximately recovered parameters.

AB - In this paper, we propose a low-rank federated learning (FL) algorithm based on singular value decomposition (SVD). The SVD factorizes the global parameters that need to be exchanged between a global server and clients for distributed model training, significantly reducing the associated communication cost. Experiment results confirm that the number of transmissions is significantly reduced while maintaining the accuracy performance of the local model using the approximately recovered parameters.

KW - edge network

KW - federated learning

KW - low-rank matrix

KW - singular value decomposition

UR - http://www.scopus.com/inward/record.url?scp=85139624057&partnerID=8YFLogxK

U2 - 10.1145/3492866.3561258

DO - 10.1145/3492866.3561258

M3 - Conference contribution

AN - SCOPUS:85139624057

T3 - Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)

SP - 285

EP - 286

BT - MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing

PB - Association for Computing Machinery

T2 - 23rd ACM International Symposium on Mobile Ad Hoc Networking and Computing, MobiHoc 2022

Y2 - 17 October 2022 through 20 October 2022

ER -

Kwon J, Park H. Efficient low-rank federated learning based on singular value decomposition. In MobiHoc 2022 - Proceedings of the 2022 23rd International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing. Association for Computing Machinery. 2022. p. 285-286. (Proceedings of the International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc)). doi: 10.1145/3492866.3561258

Efficient low-rank federated learning based on singular value decomposition

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Fingerprint

Cite this