FedSA-GCL: A Semi-Asynchronous Federated Graph Learning Framework with Personalized Aggregation and Cluster-Aware Broadcasting

2025-07-24

AI Summary

Existing federated graph learning (FGL) approaches predominantly rely on synchronous communication, suffering from low efficiency and deployment difficulty; meanwhile, mainstream asynchronous federated learning (AFL) methods neglect graph topology modeling, leading to semantic drift and representation inconsistency. To address these issues, we propose Semi-AsyncFGL, a semi-asynchronous FGL framework featuring the novel ClusterCast mechanism—enabling cluster-aware broadcasting and topology-aware personalized aggregation. Semi-AsyncFGL integrates Louvain or METIS graph partitioning to jointly mitigate data and structural heterogeneity across clients. Extensive experiments on multiple real-world graph datasets demonstrate that Semi-AsyncFGL achieves average accuracy improvements of 2.92% (Louvain) and 3.40% (METIS) over nine state-of-the-art baselines, while significantly accelerating convergence. The framework exhibits superior efficiency, robustness, and generalizability under heterogeneous graph settings.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Federated Graph Learning (FGL) is a distributed learning paradigm that enables collaborative training over large-scale subgraphs located on multiple local systems. However, most existing FGL approaches rely on synchronous communication, which leads to inefficiencies and is often impractical in real-world deployments. Meanwhile, current asynchronous federated learning (AFL) methods are primarily designed for conventional tasks such as image classification and natural language processing, without accounting for the unique topological properties of graph data. Directly applying these methods to graph learning can possibly result in semantic drift and representational inconsistency in the global model. To address these challenges, we propose FedSA-GCL, a semi-asynchronous federated framework that leverages both inter-client label distribution divergence and graph topological characteristics through a novel ClusterCast mechanism for efficient training. We evaluate FedSA-GCL on multiple real-world graph datasets using the Louvain and Metis split algorithms, and compare it against 9 baselines. Extensive experiments demonstrate that our method achieves strong robustness and outstanding efficiency, outperforming the baselines by an average of 2.92% with the Louvain and by 3.4% with the Metis.

Problem

• Addresses inefficiency in synchronous Federated Graph Learning (FGL)
• Overcomes limitations of asynchronous methods for graph data
• Prevents semantic drift in global graph learning models

Innovation

• Semi-asynchronous federated graph learning framework
• Personalized aggregation with label divergence
• ClusterCast for topology-aware broadcasting