Privacy-Preserving and Reliable Federated Learning

  • Yi Lu
  • , Lei Zhang*
  • , Lulu Wang
  • , Yuanyuan Gao
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

In Internet of Things (IoT), it is often impossible to share datasets owned by different participants (usually IoT devices) for machine learning model training due to privacy concerns. Federated learning (FL) is a promising technique to address this challenge. However, existing FL schemes face the problem of how to avoid low-quality/malicious update. To solve this problem, we propose a privacy-preserving and reliable federated learning scheme (PPRFLS) to select reliable participants and evaluate the quality of the participants’ updates. Analysis shows that the proposed scheme achieves data privacy and model reliability.

Original languageEnglish
Title of host publicationAlgorithms and Architectures for Parallel Processing - 21st International Conference, ICA3PP 2021, Proceedings
EditorsYongxuan Lai, Tian Wang, Min Jiang, Guangquan Xu, Wei Liang, Aniello Castiglione
PublisherSpringer Science and Business Media Deutschland GmbH
Pages346-361
Number of pages16
ISBN (Print)9783030953904
DOIs
StatePublished - 2022
Event21st International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2021 - Virtual, Online
Duration: 3 Dec 20215 Dec 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13157 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference21st International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2021
CityVirtual, Online
Period3/12/215/12/21

Keywords

  • Data privacy
  • Federated learning
  • Model reliability

Fingerprint

Dive into the research topics of 'Privacy-Preserving and Reliable Federated Learning'. Together they form a unique fingerprint.

Cite this