HiFGL: A Hierarchical Framework for Cross-silo Cross-device Federated Graph Learning
Zhuoning Guo, Duanyi Yao, Qiang Yang, Hao Liu
TL;DR
HiFGL tackles cross-silo cross-device federated graph learning by introducing a hierarchical three-level architecture (device-client, silo-client, server) and a Secret Message Passing scheme that preserves subgraph- and node-level privacy while maintaining graph integrity. The SecMP combines neighbor-agnostic aggregation and hierarchical Lagrangian embedding to shield sensitive information during message passing and embedding sharing, with theoretical privacy and complexity guarantees. Empirical results on Cora, CiteSeer, and PubMed show that HiFGL delivers superior graph information gain and competitive accuracy compared to state-of-the-art baselines, while preserving privacy and offering flexible applicability to cross-silo or cross-device settings. The work demonstrates that preserving graph structure across distributed, heterogeneous clients can significantly boost predictive performance in federated graphs, with practical implications for privacy-aware, scalable FL on graph data.
Abstract
Federated Graph Learning (FGL) has emerged as a promising way to learn high-quality representations from distributed graph data with privacy preservation. Despite considerable efforts have been made for FGL under either cross-device or cross-silo paradigm, how to effectively capture graph knowledge in a more complicated cross-silo cross-device environment remains an under-explored problem. However, this task is challenging because of the inherent hierarchy and heterogeneity of decentralized clients, diversified privacy constraints in different clients, and the cross-client graph integrity requirement. To this end, in this paper, we propose a Hierarchical Federated Graph Learning (HiFGL) framework for cross-silo cross-device FGL. Specifically, we devise a unified hierarchical architecture to safeguard federated GNN training on heterogeneous clients while ensuring graph integrity. Moreover, we propose a Secret Message Passing (SecMP) scheme to shield unauthorized access to subgraph-level and node-level sensitive information simultaneously. Theoretical analysis proves that HiFGL achieves multi-level privacy preservation with complexity guarantees. Extensive experiments on real-world datasets validate the superiority of the proposed framework against several baselines. Furthermore, HiFGL's versatile nature allows for its application in either solely cross-silo or cross-device settings, further broadening its utility in real-world FGL applications.