Go to ICML 2025 Conference homepage$S^2$FGL: Spatial Spectral Federated Graph Learning
Abstract: Federated Graph Learning (FGL) combines the privacy-preserving capabilities of Federated Learning (FL) with the strong graph modeling capability of Graph Neural Networks (GNNs). Current research addresses subgraph-FL from the structural perspective, neglecting the propagation of graph signals on the spatial and spectral domains of the structure. From a spatial perspective, subgraph-FL introduces edge disconnections between clients, leading to disruptions in label signals and a degradation in the semantic knowledge of the global GNN. From a spectral perspective, spectral heterogeneity causes inconsistencies in signal frequencies across subgraphs, which makes local GNNs overfit the local signal propagation schemes. As a result, spectral client drift occurs, undermining global generalizability. To tackle the challenges, we propose a global knowledge repository to mitigate the challenge of poor semantic knowledge caused by label signal disruption. Furthermore, we design a frequency alignment to address spectral client drift. The combination of **S**patial and **S**pectral strategies forms our framework $S^2$FGL. Extensive experiments on multiple datasets demonstrate the superiority of $S^2$FGL. The code is available at https://github.com/Wonder7racer/S2FGL.git.
Lay Summary: Federated Graph Learning (FGL) is a method that blends the privacy protection of federated learning with the power of Graph Neural Networks, which are great at understanding complex relationships, like mapping connections in a social network. To keep data private, FGL splits a large graph into smaller subgraphs that different devices work on separately. However, this splitting creates challenges: it breaks important semantic connections, and each device processes signals differently, causing spectral confusion. Our $S^2$FGL framework solves these issues by creating a shared "knowledge bank" to keep all devices informed about the whole network and aligning their signal frequencies to work in harmony. Tests on various datasets prove that $S^2$FGL performs better than other approaches, making it an effective way to analyze complex graphs while keeping data private.
Link To Code: https://github.com/Wonder7racer/S2FGL.git.
Primary Area: Social Aspects->Safety
Keywords: Federated Learning, Safety in Machine Learning
Submission Number: 1500
Loading