Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning

Sangwon Seo; Jaewook Lee; Haneul Ko; Sangheon Pack

doi:10.1109/WCNC51071.2022.9771600

Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning

Sangwon Seo
, Jaewook Lee
, Haneul Ko
, Sangheon Pack

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

6 Citations (Scopus)

Abstract

In federated learning (FL), which clients are selected and which quantization levels are chosen for the deep model parameters have significant impacts on the learning time as well as the learning accuracy. In this paper, we formulate a joint optimization problem on the client and quantization level selections. As a low complexity solution to the formulated problem, we develop a performance-aware client and quantization level selection (PA-CQLS) algorithm where the FL server estimates the individual round times of clients based on their computing power and channel quality, and determines the most appropriate clients and quantization levels accordingly. Simulation results show that PA-CQLS can reduce the round time by up to 70% compared to conventional algorithms.

Original language	English
Title of host publication	2022 IEEE Wireless Communications and Networking Conference, WCNC 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	1892-1897
Number of pages	6
ISBN (Electronic)	9781665442664
DOIs	https://doi.org/10.1109/WCNC51071.2022.9771600
Publication status	Published - 2022
Event	2022 IEEE Wireless Communications and Networking Conference, WCNC 2022 - Austin, United States Duration: 2022 Apr 10 → 2022 Apr 13

Publication series

Name	IEEE Wireless Communications and Networking Conference, WCNC
Volume	2022-April
ISSN (Print)	1525-3511

Conference

Conference	2022 IEEE Wireless Communications and Networking Conference, WCNC 2022
Country/Territory	United States
City	Austin
Period	22/4/10 → 22/4/13

Bibliographical note

Publisher Copyright:
© 2022 IEEE.

ASJC Scopus subject areas

General Engineering

Access to Document

10.1109/WCNC51071.2022.9771600

Cite this

Seo, S., Lee, J., Ko, H., & Pack, S. (2022). Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning. In 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022 (pp. 1892-1897). (IEEE Wireless Communications and Networking Conference, WCNC; Vol. 2022-April). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/WCNC51071.2022.9771600

Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning. / Seo, Sangwon; Lee, Jaewook; Ko, Haneul et al.
2022 IEEE Wireless Communications and Networking Conference, WCNC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. p. 1892-1897 (IEEE Wireless Communications and Networking Conference, WCNC; Vol. 2022-April).

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

Seo, S, Lee, J, Ko, H & Pack, S 2022, Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning. in 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022. IEEE Wireless Communications and Networking Conference, WCNC, vol. 2022-April, Institute of Electrical and Electronics Engineers Inc., pp. 1892-1897, 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022, Austin, United States, 22/4/10. https://doi.org/10.1109/WCNC51071.2022.9771600

Seo S, Lee J, Ko H, Pack S. Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning. In 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022. Institute of Electrical and Electronics Engineers Inc. 2022. p. 1892-1897. (IEEE Wireless Communications and Networking Conference, WCNC). doi: 10.1109/WCNC51071.2022.9771600

@inproceedings{7a0a1c1184f4401289669cbd1e218146,

title = "Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning",

abstract = "In federated learning (FL), which clients are selected and which quantization levels are chosen for the deep model parameters have significant impacts on the learning time as well as the learning accuracy. In this paper, we formulate a joint optimization problem on the client and quantization level selections. As a low complexity solution to the formulated problem, we develop a performance-aware client and quantization level selection (PA-CQLS) algorithm where the FL server estimates the individual round times of clients based on their computing power and channel quality, and determines the most appropriate clients and quantization levels accordingly. Simulation results show that PA-CQLS can reduce the round time by up to 70\% compared to conventional algorithms.",

author = "Sangwon Seo and Jaewook Lee and Haneul Ko and Sangheon Pack",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022 ; Conference date: 10-04-2022 Through 13-04-2022",

year = "2022",

doi = "10.1109/WCNC51071.2022.9771600",

language = "English",

series = "IEEE Wireless Communications and Networking Conference, WCNC",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "1892--1897",

booktitle = "2022 IEEE Wireless Communications and Networking Conference, WCNC 2022",

}

TY - GEN

T1 - Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning

AU - Seo, Sangwon

AU - Lee, Jaewook

AU - Ko, Haneul

AU - Pack, Sangheon

PY - 2022

Y1 - 2022

N2 - In federated learning (FL), which clients are selected and which quantization levels are chosen for the deep model parameters have significant impacts on the learning time as well as the learning accuracy. In this paper, we formulate a joint optimization problem on the client and quantization level selections. As a low complexity solution to the formulated problem, we develop a performance-aware client and quantization level selection (PA-CQLS) algorithm where the FL server estimates the individual round times of clients based on their computing power and channel quality, and determines the most appropriate clients and quantization levels accordingly. Simulation results show that PA-CQLS can reduce the round time by up to 70% compared to conventional algorithms.

AB - In federated learning (FL), which clients are selected and which quantization levels are chosen for the deep model parameters have significant impacts on the learning time as well as the learning accuracy. In this paper, we formulate a joint optimization problem on the client and quantization level selections. As a low complexity solution to the formulated problem, we develop a performance-aware client and quantization level selection (PA-CQLS) algorithm where the FL server estimates the individual round times of clients based on their computing power and channel quality, and determines the most appropriate clients and quantization levels accordingly. Simulation results show that PA-CQLS can reduce the round time by up to 70% compared to conventional algorithms.

UR - https://www.scopus.com/pages/publications/85130751466

U2 - 10.1109/WCNC51071.2022.9771600

DO - 10.1109/WCNC51071.2022.9771600

M3 - Conference contribution

AN - SCOPUS:85130751466

T3 - IEEE Wireless Communications and Networking Conference, WCNC

SP - 1892

EP - 1897

BT - 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2022 IEEE Wireless Communications and Networking Conference, WCNC 2022

Y2 - 10 April 2022 through 13 April 2022

ER -

Performance-Aware Client and Quantization Level Selection Algorithm for Fast Federated Learning

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this