GHOST: Generalizable One-Shot Federated Graph Learning with Proxy-Based Topology Knowledge Retention

Jiaru Qian, Guancheng Wan, Wenke Huang, Guibin Zhang, Yuxin Wu, Bo Du, Mang Ye

Abstract: Federated Graph Learning (FGL) proposes an effective approach to collaboratively training Graph Neural Networks (GNNs) while maintaining privacy. Nevertheless, communication efficiency becomes a critical bottleneck in environments with limited resources. In this context, one-shot FGL emerges as a promising solution by restricting communication to a single round. However, prevailing FGL methods face two key challenges in the one-shot setting: 1) They heavily rely on gradual personalized optimization over multiple rounds, undermining the capability of the global model to efficiently generalize across diverse graph structures. 2) They are prone to overfitting to local data distributions due to extreme structural bias, leading to catastrophic forgetting. To address these issues, we introduce **GHOST**, an innovative one-shot FGL framework. In GHOST, we establish a proxy model for each client to leverage diverse local knowledge and integrate it to train the global model. During training, we identify and consolidate parameters essential for capturing topological knowledge, thereby mitigating catastrophic forgetting. Extensive experiments on real-world tasks demonstrate the superiority and generalization capability of GHOST. The code is available at https://github.com/JiaruQian/GHOST.

Lay Summary: Federated Graph Learning (FGL) enables collaborative training of graph-based models without sharing private data, but often suffers from high communication costs. To address this, we propose **GHOST**, a one-shot FGL framework that requires only a single round of communication. GHOST introduces proxy models to capture diverse local knowledge and uses a Topology-Consistency Criterion to retain critical structural information, preventing overfitting and forgetting. Experiments on real-world tasks show that GHOST achieves strong generalization and outperforms existing methods.

Primary Area: Social Aspects->Privacy

Keywords: Federated Learning, Graph Learning

Submission Number: 4443