Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

S-VOTE: Similarity-based Voting for Client Selection in Decentralized Federated Learning

Pedro Miguel Sánchez Sánchez, Enrique Tomás Martínez Beltrán, Chao Feng, Gérôme Bovet, Gregorio Martínez Pérez, Alberto Huertas Celdrán

TL;DR

The paper tackles decentralized federated learning under non-IID data and resource constraints by introducing S-VOTE, a similarity-based voting mechanism for client selection combined with conditional local training. It leverages model similarity via cosine distance to select highly aligned clients and uses a voting protocol with adaptive random training to balance participation, reducing unnecessary communications and energy use. Empirical results on MNIST, FashionMNIST, EMNIST, and CIFAR-10 under Dirichlet non-IID partitions show S-VOTE achieves up to 21% lower communication costs, 4-6% faster convergence, and 9-17% accuracy gains in some configurations, with 14-24% energy savings. The approach demonstrates robust efficiency and generalization in heterogeneous, decentralized settings and offers practical benefits for scalable privacy-preserving learning.

Abstract

Decentralized Federated Learning (DFL) enables collaborative, privacy-preserving model training without relying on a central server. This decentralized approach reduces bottlenecks and eliminates single points of failure, enhancing scalability and resilience. However, DFL also introduces challenges such as suboptimal models with non-IID data distributions, increased communication overhead, and resource usage. Thus, this work proposes S-VOTE, a voting-based client selection mechanism that optimizes resource usage and enhances model performance in federations with non-IID data conditions. S-VOTE considers an adaptive strategy for spontaneous local training that addresses participation imbalance, allowing underutilized clients to contribute without significantly increasing resource costs. Extensive experiments on benchmark datasets demonstrate the S-VOTE effectiveness. More in detail, it achieves lower communication costs by up to 21%, 4-6% faster convergence, and improves local performance by 9-17% compared to baseline methods in some configurations, all while achieving a 14-24% energy consumption reduction. These results highlight the potential of S-VOTE to address DFL challenges in heterogeneous environments.

S-VOTE: Similarity-based Voting for Client Selection in Decentralized Federated Learning

TL;DR

The paper tackles decentralized federated learning under non-IID data and resource constraints by introducing S-VOTE, a similarity-based voting mechanism for client selection combined with conditional local training. It leverages model similarity via cosine distance to select highly aligned clients and uses a voting protocol with adaptive random training to balance participation, reducing unnecessary communications and energy use. Empirical results on MNIST, FashionMNIST, EMNIST, and CIFAR-10 under Dirichlet non-IID partitions show S-VOTE achieves up to 21% lower communication costs, 4-6% faster convergence, and 9-17% accuracy gains in some configurations, with 14-24% energy savings. The approach demonstrates robust efficiency and generalization in heterogeneous, decentralized settings and offers practical benefits for scalable privacy-preserving learning.

Abstract

Decentralized Federated Learning (DFL) enables collaborative, privacy-preserving model training without relying on a central server. This decentralized approach reduces bottlenecks and eliminates single points of failure, enhancing scalability and resilience. However, DFL also introduces challenges such as suboptimal models with non-IID data distributions, increased communication overhead, and resource usage. Thus, this work proposes S-VOTE, a voting-based client selection mechanism that optimizes resource usage and enhances model performance in federations with non-IID data conditions. S-VOTE considers an adaptive strategy for spontaneous local training that addresses participation imbalance, allowing underutilized clients to contribute without significantly increasing resource costs. Extensive experiments on benchmark datasets demonstrate the S-VOTE effectiveness. More in detail, it achieves lower communication costs by up to 21%, 4-6% faster convergence, and improves local performance by 9-17% compared to baseline methods in some configurations, all while achieving a 14-24% energy consumption reduction. These results highlight the potential of S-VOTE to address DFL challenges in heterogeneous environments.

Paper Structure

This paper contains 15 sections, 6 equations, 4 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. S-VOTE Design and Functionality
  4. Initial Federated Training
  5. Local Model Divergence
  6. Model Sharing and Similarity Computation
  7. Client Selection
  8. Voting Mechanism
  9. Conditional Local Training
  10. S-VOTE Evaluation
  11. Validation Scenario
  12. Results and Discussion
  13. Model performance
  14. Convergence Time, Energy, and Communications
  15. Conclusion and Future Work

Figures (4)

  • Figure 1: F1-Score Comparison Between the Proposed Mechanism (S-VOTE) and SoTA Aggregation Algorithms using MNIST
  • Figure 2: F1-Score Comparison Between the Proposed Mechanism (S-VOTE) and SoTA Aggregation Algorithms using FashionMNIST
  • Figure 3: F1-Score Comparison Between the Proposed Mechanism (S-VOTE) and SoTA Aggregation Algorithms using EMNIST
  • Figure 4: F1-Score Comparison Between the Proposed Mechanism (S-VOTE) and SoTA Aggregation Algorithms using CIFAR10