FedVCPL-Diff: A Federated Convolutional Prototype Learning Framework with a Diffusion Model for Speech Emotion Recognition

doi:10.1016/j.inffus.2025.103745

FedVCPL-Diff: A Federated Convolutional Prototype Learning Framework with a Diffusion Model for Speech Emotion Recognition

2025-09-16

0

PRE

AI

DOI：10.1016/j.inffus.2025.103745

Original

Original request for help

Save

Abstract

En 中文

• Federated learning enables speech emotion recognition without raw data sharing. • The approach ensures GDPR compliance, safeguarding speech data privacy and security. • Generative models are featured for data fusion of different models. • A valence-arousal-dominance exchange reduces the client communication overhead.

FedVCPL-Diff: A Federated Convolutional Prototype Learning Framework with a Diffusion Model for Speech Emotion Recognition

Abstract

Journal

Organization