FedPDM: Representation enhanced federated learning with privacy preserving diffusion models

Wei Guo

doi:10.1016/j.knosys.2026.115452

FedPDM: Representation enhanced federated learning with privacy preserving diffusion models

2026-02-02

0

PRE

AI

DOI：10.1016/j.knosys.2026.115452

Original

Original request for help

Save

Abstract

En 中文

• A diffusion-based semi-parameter-sharing federated learning framework named FedPDM. • A feature-level penalty mechanism improves the privacy-utility trade-off. • A theoretical convergence analysis is established for semi-parameter-sharing FL.

Keywords:

FedPDM

semi-parameter-sharing

federated learning

privacy-preserving

diffusion models

FedPDM: Representation enhanced federated learning with privacy preserving diffusion models

Abstract

Journal

Organization