Towards Communication-Efficient Decentralized Federated Graph Learning over Non-IID Data

2025-09-10

AI Summary

To address high communication overhead of node embeddings and performance degradation caused by coupled optimization of topology and sampling strategies in decentralized federated graph learning under non-IID data, this paper proposes Duplex—a unified adaptive framework. Its core innovation lies in the first joint modeling and optimization of communication topology structure and graph neighbor sampling ratios, enabled by a learning-driven, peer-to-peer dynamic adjustment mechanism that concurrently mitigates statistical heterogeneity and network dynamics. Extensive experiments demonstrate that, under identical resource constraints, Duplex achieves average accuracy gains of 3.3%–7.9%, training speedups of 20.1%–48.8%, and communication cost reductions of 16.7%–37.6% over state-of-the-art baselines—striking a superior balance between communication efficiency and model performance.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Decentralized Federated Graph Learning (DFGL) overcomes potential bottlenecks of the parameter server in FGL by establishing a peer-to-peer (P2P) communication network among workers. However, while extensive cross-worker communication of graph node embeddings is crucial for DFGL training, it introduces substantial communication costs. Most existing works typically construct sparse network topologies or utilize graph neighbor sampling methods to alleviate the communication overhead in DFGL. Intuitively, integrating these methods may offer promise for doubly improving communication efficiency in DFGL. However, our preliminary experiments indicate that directly combining these methods leads to significant training performance degradation if they are jointly optimized. To address this issue, we propose Duplex, a unified framework that jointly optimizes network topology and graph sampling by accounting for their coupled relationship, thereby significantly reducing communication cost while enhancing training performance in DFGL. To overcome practical DFGL challenges, eg, statistical heterogeneity and dynamic network environments, Duplex introduces a learning-driven algorithm to adaptively determine optimal network topologies and graph sampling ratios for workers. Experimental results demonstrate that Duplex reduces completion time by 20.1%--48.8% and communication costs by 16.7%--37.6% to achieve target accuracy, while improving accuracy by 3.3%--7.9% under identical resource budgets compared to baselines.

Problem

• Reducing communication costs in decentralized federated graph learning
• Overcoming performance degradation from combined optimization methods
• Addressing statistical heterogeneity and dynamic network challenges

Innovation

• Jointly optimizes network topology and graph sampling
• Uses learning-driven algorithm for adaptive optimization
• Reduces communication costs while enhancing training performance