pFedGame -- Decentralized Federated Learning using Game Theory in Dynamic Topology

TL;DR

This paper introduces pFedGame, a decentralized federated learning approach using game theory that operates without a central server, effectively handling dynamic network topologies and improving convergence and accuracy.

Contribution

It presents a novel decentralized federated learning algorithm based on game theory that addresses issues of centralization, data heterogeneity, and dynamic network topology.

Findings

Achieves over 70% accuracy on heterogeneous data

Outperforms existing decentralized federated learning methods

Effectively handles dynamic network topologies

Abstract

Conventional federated learning frameworks suffer from several challenges including performance bottlenecks at the central aggregation server, data bias, poor model convergence, and exposure to model poisoning attacks, and limited trust in the centralized infrastructure. In the current paper, a novel game theory-based approach called pFedGame is proposed for decentralized federated learning, best suitable for temporally dynamic networks. The proposed algorithm works without any centralized server for aggregation and incorporates the problem of vanishing gradients and poor convergence over temporally dynamic topology among federated learning participants. The solution comprises two sequential steps in every federated learning round, for every participant. First, it selects suitable peers for collaboration in federated learning. Secondly, it executes a two-player constant sum cooperative game to reach convergence by applying an optimal federated learning aggregation strategy. Experiments performed to assess the performance of pFedGame in comparison to existing methods in decentralized federated learning have shown promising results with accuracy higher than 70% for heterogeneous data.