FedGCN: Convergence and Communication Tradeoffs in Federated Training of Graph Convolutional Networks

Distributed methods for training models on graph datasets have recently grown in popularity, due to the size of graph datasets as well as the private nature of graphical data like social networks. However, the graphical structure of this data means that it cannot be disjointly partitioned between different learning clients, leading to either significant communication overhead between clients or a loss of information available to the training method. We introduce Federated Graph Convolutional Network (FedGCN), which uses federated learning to train GCN models with optimized convergence rate and communication cost. Compared to prior methods that require communication among clients at each iteration, FedGCN preserves the privacy of client data and only needs communication at the initial step, which greatly reduces communication cost and speeds up the convergence rate. We theoretically analyze the tradeoff between FedGCN's convergence rate and communication cost under different data distributions, introducing a general framework can be generally used for the analysis of all edge-completion-based GCN training algorithms. Experimental results demonstrate the effectiveness of our algorithm and validate our theoretical analysis.