GL-GNN: Graph learning via the network of graphs

Yixiang Shan, Jielong Yang, Yixing Gao

Abstract: Highlights•We propose a novel network-of-graphs based graph-learnable method GL-GNN.•GL-GNN is the first model that uses the network of graphs to learn graph structures for GNNs.•GL-GNN is more robust to edge attack than multiple recently proposed methods.•GL-GNN outperforms 12 baseline methods on six out of seven datasets without available graphs.•GL-GNN outperforms another 16 out of 17 baseline methods on accuracy and F1 score on seven datasets with graphs available.