Low Complexity Byzantine-Resilient Federated Learning

<p dir="ltr">Federated learning (FL) has gained attention for enabling efficient distributed learning while maintaining data privacy. However, the data privacy constraint reduces the transparency in the agents’ model update making the learning process vulnerable to Byzantine attacks....

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: A. Gouissem (17541396) (author)
مؤلفون آخرون: S. Hassanein (21399926) (author), K. Abualsaud (17541399) (author), E. Yaacoub (17541402) (author), M. Mabrok (21399929) (author), M. Abdallah (812014) (author), T. Khattab (17541405) (author), M. Guizani (17541408) (author)
منشور في: 2024
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513543182745600
author A. Gouissem (17541396)
author2 S. Hassanein (21399926)
K. Abualsaud (17541399)
E. Yaacoub (17541402)
M. Mabrok (21399929)
M. Abdallah (812014)
T. Khattab (17541405)
M. Guizani (17541408)
author2_role author
author
author
author
author
author
author
author_facet A. Gouissem (17541396)
S. Hassanein (21399926)
K. Abualsaud (17541399)
E. Yaacoub (17541402)
M. Mabrok (21399929)
M. Abdallah (812014)
T. Khattab (17541405)
M. Guizani (17541408)
author_role author
dc.creator.none.fl_str_mv A. Gouissem (17541396)
S. Hassanein (21399926)
K. Abualsaud (17541399)
E. Yaacoub (17541402)
M. Mabrok (21399929)
M. Abdallah (812014)
T. Khattab (17541405)
M. Guizani (17541408)
dc.date.none.fl_str_mv 2024-11-28T03:00:00Z
dc.identifier.none.fl_str_mv 10.1109/tifs.2024.3482727
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Low_Complexity_Byzantine-Resilient_Federated_Learning/29605283
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Artificial intelligence
Distributed computing and systems software
Machine learning
Mathematical sciences
Numerical and computational mathematics
Statistics
Federated learning
Distributed learning
Byzantine attacks
Convergence analysis
dc.title.none.fl_str_mv Low Complexity Byzantine-Resilient Federated Learning
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p dir="ltr">Federated learning (FL) has gained attention for enabling efficient distributed learning while maintaining data privacy. However, the data privacy constraint reduces the transparency in the agents’ model update making the learning process vulnerable to Byzantine attacks. In this paper, a mathematical proof is provided to show that when the traditional model-combining scheme is used, the model will eventually diverge to non-useful solutions in the presence of Byzantine agents independently from their number or their contributions. A low complexity norm-control based aggregation approach is also proposed and shown to converge to the optimal and sub-optimal solutions in the absence or presence of Byzantine nodes, respectively. Monte-Carlo simulations are also conducted to verify and validate the mathematical derivations and the efficiency of the proposed approach in protecting the FL model.</p><h2>Other Information</h2><p dir="ltr">Published in: IEEE Transactions on Information Forensics and Security<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" rel="noreferrer noopener" target="_blank">https://creativecommons.org/licenses/by/4.0/</a>  <br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/tifs.2024.3482727" target="_blank">https://dx.doi.org/10.1109/tifs.2024.3482727</a></p>
eu_rights_str_mv openAccess
id Manara2_1634166746de32beaa6953e1b6023aae
identifier_str_mv 10.1109/tifs.2024.3482727
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/29605283
publishDate 2024
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Low Complexity Byzantine-Resilient Federated LearningA. Gouissem (17541396)S. Hassanein (21399926)K. Abualsaud (17541399)E. Yaacoub (17541402)M. Mabrok (21399929)M. Abdallah (812014)T. Khattab (17541405)M. Guizani (17541408)Information and computing sciencesArtificial intelligenceDistributed computing and systems softwareMachine learningMathematical sciencesNumerical and computational mathematicsStatisticsFederated learningDistributed learningByzantine attacksConvergence analysis<p dir="ltr">Federated learning (FL) has gained attention for enabling efficient distributed learning while maintaining data privacy. However, the data privacy constraint reduces the transparency in the agents’ model update making the learning process vulnerable to Byzantine attacks. In this paper, a mathematical proof is provided to show that when the traditional model-combining scheme is used, the model will eventually diverge to non-useful solutions in the presence of Byzantine agents independently from their number or their contributions. A low complexity norm-control based aggregation approach is also proposed and shown to converge to the optimal and sub-optimal solutions in the absence or presence of Byzantine nodes, respectively. Monte-Carlo simulations are also conducted to verify and validate the mathematical derivations and the efficiency of the proposed approach in protecting the FL model.</p><h2>Other Information</h2><p dir="ltr">Published in: IEEE Transactions on Information Forensics and Security<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" rel="noreferrer noopener" target="_blank">https://creativecommons.org/licenses/by/4.0/</a>  <br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/tifs.2024.3482727" target="_blank">https://dx.doi.org/10.1109/tifs.2024.3482727</a></p>2024-11-28T03:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1109/tifs.2024.3482727https://figshare.com/articles/journal_contribution/Low_Complexity_Byzantine-Resilient_Federated_Learning/29605283CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/296052832024-11-28T03:00:00Z
spellingShingle Low Complexity Byzantine-Resilient Federated Learning
A. Gouissem (17541396)
Information and computing sciences
Artificial intelligence
Distributed computing and systems software
Machine learning
Mathematical sciences
Numerical and computational mathematics
Statistics
Federated learning
Distributed learning
Byzantine attacks
Convergence analysis
status_str publishedVersion
title Low Complexity Byzantine-Resilient Federated Learning
title_full Low Complexity Byzantine-Resilient Federated Learning
title_fullStr Low Complexity Byzantine-Resilient Federated Learning
title_full_unstemmed Low Complexity Byzantine-Resilient Federated Learning
title_short Low Complexity Byzantine-Resilient Federated Learning
title_sort Low Complexity Byzantine-Resilient Federated Learning
topic Information and computing sciences
Artificial intelligence
Distributed computing and systems software
Machine learning
Mathematical sciences
Numerical and computational mathematics
Statistics
Federated learning
Distributed learning
Byzantine attacks
Convergence analysis