Vertically Federated Graph Neural Network for Privacy-Preserving Node Classification

• DOI: 10.48550/arxiv.2005.11903
• Publication Date: 2020-01-01
• Authors: Chaochao Chen; Jun Zhou; Longfei Zheng; Huiwen Wu; Lingjuan Lyu; Jia Wu; Bingzhe Wu; Ziqi Liu; Li Wang; Xiaolin Zheng
• Keywords: Delegate; Differential Privacy; Information leakage

Recently, Graph Neural Network (GNN) has achieved remarkable progresses in various real-world tasks on graph data, consisting of node features and the adjacent information between different nodes. High-performance GNN models always depend both rich complete edge graph. However, such could possibly be isolated by data holders practice, which is so-called isolation problem. To solve this problem, paper, we propose VFGNN, a federated learning paradigm for privacy-preserving classification task under vertically partitioned setting, can generalized to existing models. Specifically, split computation into two parts. We leave private (i.e., features, edges, labels) related computations holders, delegate rest semi-honest server. also apply differential privacy prevent potential leakage from conduct experiments three benchmarks results demonstrate effectiveness VFGNN.