Decentralized Personalized Federated Learning

TL;DR

This paper introduces DPFL, a novel decentralized federated learning method that constructs personalized collaboration graphs using a bi-level optimization framework, significantly improving resource efficiency and model personalization in heterogeneous data environments.

Contribution

The work presents a new communication-efficient, graph-based personalized federated learning approach that considers client relations at a granular level, outperforming existing methods.

Findings

DPFL outperforms baseline methods across diverse datasets.

It effectively handles data heterogeneity and reduces communication costs.

The approach enhances resource efficiency and personalization in decentralized settings.

Abstract

This work tackles the challenges of data heterogeneity and communication limitations in decentralized federated learning. We focus on creating a collaboration graph that guides each client in selecting suitable collaborators for training personalized models that leverage their local data effectively. Our approach addresses these issues through a novel, communication-efficient strategy that enhances resource efficiency. Unlike traditional methods, our formulation identifies collaborators at a granular level by considering combinatorial relations of clients, enhancing personalization while minimizing communication overhead. We achieve this through a bi-level optimization framework that employs a constrained greedy algorithm, resulting in a resource-efficient collaboration graph for personalized learning. Extensive evaluation against various baselines across diverse datasets demonstrates the superiority of our method, named DPFL. DPFL consistently outperforms other approaches, showcasing its effectiveness in handling real-world data heterogeneity, minimizing communication overhead, enhancing resource efficiency, and building personalized models in decentralized federated learning scenarios.