Abstract: In this paper, I explore the application of block Krylov subspace methods to enhance the scalability and efficiency of deep graph convolutional networks (GCNs). Deep learning models, ...