SGCN: A scalable graph convolutional network with graph-shaped kernels and multi-channels

Zhenhua Huang, Wenhao Zhou, Kunhao Li, Zhaohong Jia

Abstract: Highlights•We introduce a novel GNN architecture that utilizes flexible graph-shaped convolution kernels and multi-channels.•Our model outperforms existing GNNs in node classification across multiple graph datasets.•The proposed model is capable of handling graphs of different scales.