Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Shiwei Lu, Yuhang He, Jiashuo Li, Qiang Wang, Yihong Gong

TL;DR

A novel Generative Federated Prototype Learning framework that improves model accuracy by 3.6% under imbalanced data settings while maintaining low communication cost, and devise a dual-classifier architecture, optimized via a hybrid loss combining Dot Regression and Cross-Entropy.

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.

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

TL;DR

A novel Generative Federated Prototype Learning framework that improves model accuracy by 3.6% under imbalanced data settings while maintaining low communication cost, and devise a dual-classifier architecture, optimized via a hybrid loss combining Dot Regression and Cross-Entropy.

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.
Paper Structure (31 sections, 1 theorem, 28 equations, 7 figures, 3 tables, 2 algorithms)

This paper contains 31 sections, 1 theorem, 28 equations, 7 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. The knowledge fusion of imbalanced data in FL
  4. Prototype learning
  5. Generative learning
  6. Problem setting
  7. Knowledge fusion of imbalanced data in federated learning
  8. Classifier with equiangular tight frame (ETF)
  9. Gaussian Mixture Model (GMM)
  10. Bhattacharyya Distance
  11. Methodology
  12. Feature Alignment with the Dual-Classifier Structure (DCS)
  13. Improving model generalization with Pseudo Feature Generation (PFG)
  14. Experiments
  15. Experiments setup
  16. ...and 16 more sections

Key Result

Theorem A.1

Under Assumptions 1-4, if the learning rate $\eta$ satisfies $\eta \leq \frac{1}{L}$, then after $T$ communication rounds, the sequence of parameters $\{\Theta^t\}$ generated by the GFPL algorithm satisfies: where $C_1$ and $C_2$ are positive constants.

Figures (7)

  • Figure 1: Generative Federated Prototype Learning (GFPL):(a) the client trains feature extractor, projection layers through local data, while generating local prototypes with GMM and uploading them to the server for prototype interaction; (b) After receiving global prototypes, the client replaces local prototypes with global prototypes and further generates balanced pseudo features through GMM sampling, thereby retraining the model projection layer.
  • Figure 2: Visualization of features and projections under different loss functions: (a) only $\mathcal{L}_{DR}$ (features); (b) $\mathcal{L}_{DR}+\mathcal{L}_{CE}$ (features); (c) only $\mathcal{L}_{DR}$ (projections); (d) $\mathcal{L}_{DR}+\mathcal{L}_{CE}$ (projections)
  • Figure 3: The influence of hyperparameters on model performance: (a) $\lambda$; (b) GMM commponent number; (c)The average of shot.
  • Figure 4: The influence of hyperparameters on model performance: (a)$\hat{w}$; (b)Retraining interval; (c)Initial round of prototype interaction
  • Figure 5: Error bars of test accuracy on different datasets: (a-c) MNIST dataset with different $\bar{w}$ values; (d-f) FEMNIST dataset with different $\bar{w}$ values; (g-i) CIFAR10 dataset with different $\bar{w}$.
  • ...and 2 more figures

Theorems & Definitions (2)

  • Theorem A.1
  • proof