HYPERION: Fine-Grained Hypersphere Alignment for Robust Federated Graph Learning

Guancheng Wan (University of California, Los Angeles) · Wenke Huang (Wuhan University) · Guibin Zhang (International Digital Economy Academy, International Digital Economy Academy) · Bo Du (Wuhan University) · Yue Liu (NUS) · Yanbiao Ma (Xidian University) · Yuxin Wu (Renmin University of China) · Xiaoran Shang (Wuhan University) · Jinhe Bi (Ludwig-Maximilians-Universität München) · Liangtao Zheng (University of California, San Diego) · Xin Lin (University of California, San Diego)
cora datasetdecentralized frameworkedge-mediated error propagationf1-macro scorefine-grained hypersphere alignmentgeometric-aware hyperspherical purification modulegraph neural networksgraph structureshyperspherical embeddingintra-class topological differenceslocalized perspective constraintsnoisy-label environmentsrobust federated graph learningrobustnessunified hyperspherical space

Robust Federated Graph Learning (FGL) provides an effective decentralized framework for training Graph Neural Networks (GNNs) in noisy-label environments. However, the subtlety of noise during training presents formidable obstacles for developing robust FGL systems. Previous robust FL approaches neither adequately constrain edge-mediated error propagation nor account for intra-class topological differences. At the client level, we innovatively demonstrate that hyperspherical embedding can effectively capture graph structures in a fine-grained manner. Correspondingly, our method effectively addresses the aforementioned issues through fine-grained hypersphere alignment. Moreover, we uncover undetected noise arising from localized perspective constraints and propose the geometric-aware hyperspherical purification module at the server level. Combining both level strategies, we present our robust FGL framework,**HYPERION**, which operates all components within a unified hyperspherical space. **HYPERION** demonstrates remarkable robustness across multiple datasets, for instance, achieving a 29.7\% $\uparrow$ F1-macro score with 50\%-pair noise on Cora. The code is available for anonymous access at \url{https://anonymous.4open.science/r/Hyperion-NeurIPS/}.