Zur Hauptnavigation wechseln Zur Suche wechseln Zum Hauptinhalt wechseln

FedP2PAvg: A Peer-to-Peer Collaborative Framework for Federated Learning in Non-IID Scenarios

  • Bruno J.T. Fernandes*
  • , Agostinho Freire
  • , João V R. de Andrade
  • , Leandro H.S. Silva
  • , Nicolás Navarro-Guerrero
  • *Korrespondierende*r Autor*in für diese Arbeit

Publikation: Beitrag in Buch/Bericht/Sammelwerk/KonferenzbandAufsatz in KonferenzbandForschungPeer-Review

Abstract

Federated learning is a decentralized machine learning approach where models are trained collaboratively across multiple devices or nodes holding local data without sharing that data directly. It enables privacy-preserving, scalable, and collaborative machine learning. One of the key challenges in federated learning is its inefficiency in handling scenarios where data is highly imbalanced and non-independent and identically distributed (non-IID) across local nodes, leading to biased global models and slow convergence. This paper introduces a peer-to-peer refinement mechanism combined with FedAvg aggregation to enhance model accuracy in highly imbalanced and non-IID federated learning scenarios. Experiments were conducted on the MNIST, Fashion-MNIST and CIFAR-10 datasets using a Dirichlet distribution with α=0.1 to simulate highly imbalanced and non-IID data scenarios. The results demonstrated that the proposed approach achieved higher accuracy, 98.17% in MNIST, 84.35% in Fashion-MNIST and 67.49% in CIFAR-10 while requiring less than half the number of rounds to converge compared to traditional federated learning methods.

OriginalspracheEnglisch
Titel des SammelwerksArtificial Neural Networks and Machine Learning – ICANN 2025 - 34th International Conference on Artificial Neural Networks, 2025, Proceedings
Herausgeber/-innenWalter Senn, Marcello Sanguineti, Ausra Saudargiene, Igor V. Tetko, Alessandro E. P. Villa, Viktor Jirsa, Yoshua Bengio
Herausgeber (Verlag)Springer Science and Business Media Deutschland GmbH
Seiten391-403
Seitenumfang13
ISBN (elektronisch)978-3-032-04558-4
ISBN (Print)9783032045577
DOIs
PublikationsstatusVeröffentlicht - 12 Sept. 2026
Veranstaltung34th International Conference on Artificial Neural Networks, ICANN 2025 - Kaunas, Litauen
Dauer: 9 Sept. 202512 Sept. 2025

Publikationsreihe

NameLecture Notes in Computer Science
Band16068 LNCS
ISSN (Print)0302-9743
ISSN (elektronisch)1611-3349

Konferenz

Konferenz34th International Conference on Artificial Neural Networks, ICANN 2025
KurztitelICANN 2025
Land/GebietLitauen
OrtKaunas
Zeitraum9 Sept. 202512 Sept. 2025

ASJC Scopus Sachgebiete

  • Theoretische Informatik
  • Allgemeine Computerwissenschaft

Dieses zitieren