Accelerating Federated Learning via Modified Local Model Update Based on Individual Performance Metric

The privacy-preserving federated learning (FL) algorithm is considered one of the most widely used distributed training algorithms. Its effectiveness is primarily observed when the datasets on the clients are independent, identically distributed (IID), and balanced. However, in real-world situations...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2023 3rd International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME) s. 1 - 6
Hlavní autoři: Barhoush, Mahdi, Ayad, Ahmad, Schmeink, Anke
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 19.07.2023
Témata:
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Abstract The privacy-preserving federated learning (FL) algorithm is considered one of the most widely used distributed training algorithms. Its effectiveness is primarily observed when the datasets on the clients are independent, identically distributed (IID), and balanced. However, in real-world situations, the datasets on the clients are often non-IID, leading to varied data and feature distributions, thereby adversely affecting the performance of federated learning. To address this challenge, this work applies a modified local model update mechanism based on the individual performance metric of all clients' models under training calculated on a reference testing dataset that resides on the server side. The proposed modification enhances each client model's performance individually, regardless of its data distribution. For instance, if a client model poorly predicts certain classes, it will receive a higher percentage and weighting factor from other models that perform better in predicting those classes. The empirical studies of the modified algorithm for FedAvg and FedProx aggregation methods under IId and non-IID data distribution with MNIST and CIFAR10 datasets show that this approach can speed up the training process, increase the overall system accuracy, and reduce the number of communication rounds.
AbstractList The privacy-preserving federated learning (FL) algorithm is considered one of the most widely used distributed training algorithms. Its effectiveness is primarily observed when the datasets on the clients are independent, identically distributed (IID), and balanced. However, in real-world situations, the datasets on the clients are often non-IID, leading to varied data and feature distributions, thereby adversely affecting the performance of federated learning. To address this challenge, this work applies a modified local model update mechanism based on the individual performance metric of all clients' models under training calculated on a reference testing dataset that resides on the server side. The proposed modification enhances each client model's performance individually, regardless of its data distribution. For instance, if a client model poorly predicts certain classes, it will receive a higher percentage and weighting factor from other models that perform better in predicting those classes. The empirical studies of the modified algorithm for FedAvg and FedProx aggregation methods under IId and non-IID data distribution with MNIST and CIFAR10 datasets show that this approach can speed up the training process, increase the overall system accuracy, and reduce the number of communication rounds.
Author Barhoush, Mahdi
Schmeink, Anke
Ayad, Ahmad
Author_xml – sequence: 1
  givenname: Mahdi
  surname: Barhoush
  fullname: Barhoush, Mahdi
  email: mahdi.barhoush@inda.rwth-aachen.de
  organization: RWTH University,Chair of Information Theory and Data Analytics (INDA),Aachen,Germany
– sequence: 2
  givenname: Ahmad
  surname: Ayad
  fullname: Ayad, Ahmad
  email: ahmad.ayad@inda.rwth-aachen.de
  organization: RWTH University,Chair of Information Theory and Data Analytics (INDA),Aachen,Germany
– sequence: 3
  givenname: Anke
  surname: Schmeink
  fullname: Schmeink, Anke
  email: schmeink@inda.rwth-aachen.de
  organization: RWTH University,Chair of Information Theory and Data Analytics (INDA),Aachen,Germany
BookMark eNo1T01LwzAYjqAHnfsHHoL31jdJuzTHWTottOjBnUf65q0EunRkdeC_t0U9PZ888Nyx6zAGYuxRQCoEmKe6rMqyrXJdKEglSJUKkLkCDVdsbbQpVA5KSqPlLeu3iDRQtJMPn3xHbqHkeEM2hsW6eMvb0fneL-6IdlgkDXx_cnOTP9vzHIyB18H5i3dfc-GdYj_Gow1IvKUperxnN70dzrT-wxXb76qP8jVp3l7qctskXggzJZh1ZKzbCMwcAuQbFEIWpHslM00CO6mlI9BKImG-BB3ZTBjtsMgJnFqxh99dT0SHU_RHG78P__fVD6UaVrg
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/ICECCME57830.2023.10253070
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Xplore POP ALL
IEEE Xplore All Conference Proceedings
IEEE Xplore Electronic Library
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Xplore Electronic Library
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
EISBN 9798350322972
EndPage 6
ExternalDocumentID 10253070
Genre orig-research
GrantInformation_xml – fundername: Ministry of Education
  funderid: 10.13039/100010002
GroupedDBID 6IE
6IL
CBEJK
RIE
RIL
ID FETCH-LOGICAL-i119t-c4be9ad61c4dc0056c1128e7f3247e1cb272de0732cec58e7fbea4197dc85e0d3
IEDL.DBID RIE
IngestDate Wed Sep 27 05:40:30 EDT 2023
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i119t-c4be9ad61c4dc0056c1128e7f3247e1cb272de0732cec58e7fbea4197dc85e0d3
PageCount 6
ParticipantIDs ieee_primary_10253070
PublicationCentury 2000
PublicationDate 2023-July-19
PublicationDateYYYYMMDD 2023-07-19
PublicationDate_xml – month: 07
  year: 2023
  text: 2023-July-19
  day: 19
PublicationDecade 2020
PublicationTitle 2023 3rd International Conference on Electrical, Computer, Communications and Mechatronics Engineering (ICECCME)
PublicationTitleAbbrev ICECCME
PublicationYear 2023
Publisher IEEE
Publisher_xml – name: IEEE
Score 1.8677948
Snippet The privacy-preserving federated learning (FL) algorithm is considered one of the most widely used distributed training algorithms. Its effectiveness is...
SourceID ieee
SourceType Publisher
StartPage 1
SubjectTerms Computational modeling
Enhance Performance
FedAvg
Federated learning
FedProx
Measurement
Mechatronics
Modified Algorithms
Prediction algorithms
Predictive models
Training
Title Accelerating Federated Learning via Modified Local Model Update Based on Individual Performance Metric
URI https://ieeexplore.ieee.org/document/10253070
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV09b8IwELVK1aFTW5Wq3_LQNTROnDge2whUBhBDkdiQfb6gSFWCKPD76wsB1KFDt9hRZMnn0zs7fu8x9mKkdTILTWASkwbSiDiwmfBbFSjiEJzTWWobswk1HmezmZ60ZPWGC4OIzeUz7NFj8y_f1bChozKf4VFCa7TDOkqlO7JWKyQqQv06zPt5Pur7NRiHPbIF7-0_-GWd0iDH4OKfY16y7pGDxycHdLliJ1hds-INwOMERa1a8AEJQfha0fFWJXXBt6Xho9qVRUm9hFPUxC8-XdLWnr970HK8rvjwQMTikyN3gI_IYAu6bDrof-YfQeuUEJRC6HUA0qI2LhUgHZC6J_gyKkNV-HJJoQAbqcihz-YIEBJ6YdFIoZWDLMHQxTfstKorvGW8SG1srQpF4YHbGG0TI32gXaSjFH3y37EuTdJ8uRPDmO_n5_6P_gd2TqGg41ChH9nperXBJ3YG23X5vXpuQvgDEx2fxw
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3PS8MwFA46BT2pOPG3OXjtbPo7Ry0bG65jhw12G8nL6yhIO-a2v9-82jk8ePDWpJRAXh7fS5rv-xh7VoE2QeIqR4UqcgIlfEcnwm5VIPddMEYmka7NJuLRKJnN5Lghq9dcGESsL59hhx7rf_mmgg0dldkM90Jao4fsiKyzGrpWIyUqXPkySLtpmnXtKvTdDhmDd3af_DJPqbGjd_bPUc9Ze8_C4-MffLlgB1hesvwVwCIFxa1c8B5JQdhq0fBGJ3XBt4XiWWWKvKBeQipq4gefLmlzz98sbBlelXzwQ8Xi4z17gGdksQVtNu11J2nfabwSnEIIuXYg0CiViQQEBkjfE2whlWCc24IpRgHaiz2DNp89QAjphUYVCBkbSEJ0jX_FWmVV4jXjeaR9rWNX5Ba6lZI6VIENtfGkF6FN_xvWpkmaL7_lMOa7-bn9o_-JnfQn2XA-HIze79gphYUOR4W8Z631aoMP7Bi26-Jz9ViH8wuTB6MQ
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=proceeding&rft.title=2023+3rd+International+Conference+on+Electrical%2C+Computer%2C+Communications+and+Mechatronics+Engineering+%28ICECCME%29&rft.atitle=Accelerating+Federated+Learning+via+Modified+Local+Model+Update+Based+on+Individual+Performance+Metric&rft.au=Barhoush%2C+Mahdi&rft.au=Ayad%2C+Ahmad&rft.au=Schmeink%2C+Anke&rft.date=2023-07-19&rft.pub=IEEE&rft.spage=1&rft.epage=6&rft_id=info:doi/10.1109%2FICECCME57830.2023.10253070&rft.externalDocID=10253070