GFPL: Generative Federated Prototype Learning for Resource-Constrained and Data-Imbalanced Vision Task

TL;DR

GFPL introduces a prototype-based federated learning framework inspired by the human brain, effectively addressing data imbalance and communication challenges in resource-constrained vision tasks, with improved accuracy and efficiency.

Contribution

The paper proposes GFPL, a novel federated learning method using Gaussian Mixture Model prototypes and a dual-classifier architecture to handle data imbalance and reduce communication overhead.

Findings

Improves accuracy by 3.6% on benchmarks with imbalanced data

Reduces communication cost compared to traditional FL methods

Effectively fuses knowledge across clients using prototype aggregation

Abstract

Federated learning (FL) facilitates the secure utilization of decentralized images, advancing applications in medical image recognition and autonomous driving. However, conventional FL faces two critical challenges in real-world deployment: ineffective knowledge fusion caused by model updates biased toward majority-class features, and prohibitive communication overhead due to frequent transmissions of high-dimensional model parameters. Inspired by the human brain's efficiency in knowledge integration, we propose a novel Generative Federated Prototype Learning (GFPL) framework to address these issues. Within this framework, a prototype generation method based on Gaussian Mixture Model (GMM) captures the statistical information of class-wise features, while a prototype aggregation strategy using Bhattacharyya distance effectively fuses semantically similar knowledge across clients. In addition, these fused prototypes are leveraged to generate pseudo-features, thereby mitigating feature distribution imbalance across clients. To further enhance feature alignment during local training, we devise a dual-classifier architecture, optimized via a hybrid loss combining Dot Regression and Cross-Entropy. Extensive experiments on benchmarks show that GFPL improves model accuracy by 3.6% under imbalanced data settings while maintaining low communication cost.