Toward Representing Identical Privacy-Preserving Graph Neural Network via Split Learning

Abstract

In recent years, the fast rise in number of studies on graph neural network (GNN) has put it from the theories research to the real-world application stage. Despite the encouraging performance achieved by GNN, less attention has been paid to the privacy-preserving training and inference over distributed graph data in the related literature. Due to the particularity of graph structure, it is challenging to extend the existing private learning frameworks to GNN. Motivated by the idea of split learning, we propose a server aided privacy-preserving GNN (SAPGNN) for the intra-graph node level task on the horizontally partitioned cross-silo scenario. It offers a natural extension of centralized GNN to the isolated graph with max/min pooling aggregation, while guaranteeing that all the private data involved in the computation still stays with local data holders. To further enhance the data privacy, a secure pooling aggregation mechanism is proposed. Theoretical and experimental results show that the proposed model achieves the same accuracy as the one learned over the combined data.

Keywords

• Graph neural networks
• message passing
• privacy-preserving
• split learning