Forget-SVGD: Particle-Based Bayesian Federated Unlearning

Jinu Gong, Joonhyuk Kang, Osvaldo Simeone, Rahif Kassab

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Variational particle-based Bayesian learning methods have the advantage of not being limited by the bias affecting more conventional parametric techniques. This paper proposes to leverage the flexibility of non-parametric Bayesian approximate inference to develop a novel Bayesian federated unlearning method, referred to as Forget-Stein Variational Gradient Descent (Forget-SVGD). Forget-SVGD builds on SVGD - a particle-based approximate Bayesian inference scheme using gradient-based deterministic updates - and on its distributed (federated) extension known as Distributed SVGD (DSVGD). Upon the completion of federated learning, as one or more participating agents request for their data to be 'forgotten', Forget-SVGD carries out local SVGD updates at the agents whose data need to be 'unlearned', which are interleaved with communication rounds with a parameter server. The proposed method is validated via performance comparisons with non-parametric schemes that train from scratch by excluding data to be forgotten, as well as with existing parametric Bayesian unlearning methods.

Original languageEnglish (US)
Title of host publication2022 IEEE Data Science and Learning Workshop, DSLW 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665454261
DOIs
StatePublished - 2022
Externally publishedYes
Event2022 IEEE Data Science and Learning Workshop, DSLW 2022 - Singapore, Singapore
Duration: May 22 2022May 23 2022

Publication series

Name2022 IEEE Data Science and Learning Workshop, DSLW 2022

Conference

Conference2022 IEEE Data Science and Learning Workshop, DSLW 2022
Country/TerritorySingapore
CitySingapore
Period5/22/225/23/22

All Science Journal Classification (ASJC) codes

  • Computer Science Applications
  • Information Systems
  • Artificial Intelligence

Keywords

  • Bayesian learning
  • Federated learning
  • Machine unlearning
  • Stein variational gradient descent

Fingerprint

Dive into the research topics of 'Forget-SVGD: Particle-Based Bayesian Federated Unlearning'. Together they form a unique fingerprint.

Cite this