Federated Unlearning: a Perspective of Stability and Fairness

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Shao, Jiaqi, Lin, Tao, Cao, Xuanyu, Luo, Bing
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917681493442560
author Shao, Jiaqi
Lin, Tao
Cao, Xuanyu
Luo, Bing
author_facet Shao, Jiaqi
Lin, Tao
Cao, Xuanyu
Luo, Bing
contents This paper explores the multifaceted consequences of federated unlearning (FU) with data heterogeneity. We introduce key metrics for FU assessment, concentrating on verification, global stability, and local fairness, and investigate the inherent trade-offs. Furthermore, we formulate the unlearning process with data heterogeneity through an optimization framework. Our key contribution lies in a comprehensive theoretical analysis of the trade-offs in FU and provides insights into data heterogeneity's impacts on FU. Leveraging these insights, we propose FU mechanisms to manage the trade-offs, guiding further development for FU mechanisms. We empirically validate that our FU mechanisms effectively balance trade-offs, confirming insights derived from our theoretical analysis.
format Preprint
id arxiv_https___arxiv_org_abs_2402_01276
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Federated Unlearning: a Perspective of Stability and Fairness
Shao, Jiaqi
Lin, Tao
Cao, Xuanyu
Luo, Bing
Artificial Intelligence
This paper explores the multifaceted consequences of federated unlearning (FU) with data heterogeneity. We introduce key metrics for FU assessment, concentrating on verification, global stability, and local fairness, and investigate the inherent trade-offs. Furthermore, we formulate the unlearning process with data heterogeneity through an optimization framework. Our key contribution lies in a comprehensive theoretical analysis of the trade-offs in FU and provides insights into data heterogeneity's impacts on FU. Leveraging these insights, we propose FU mechanisms to manage the trade-offs, guiding further development for FU mechanisms. We empirically validate that our FU mechanisms effectively balance trade-offs, confirming insights derived from our theoretical analysis.
title Federated Unlearning: a Perspective of Stability and Fairness
topic Artificial Intelligence
url https://arxiv.org/abs/2402.01276