LeDA-GNN: Learnable dual augmentation for graph neural networks

Gen Liu, Zhongying Zhao, Chao Li, Yanwei Yu

Abstract: Highlights•We propose a graph data augmentation method named LeDA-GNN.•The proposed LeDA-GNN can achieve data augmentation at the feature and topology level.•The proposed LeDA-GNN can perform data augmentation in a learnable way.•The experimental results on four datasets demonstrate the superiority of LeDA-GNN.