Information-controlled graph convolutional network for multi-view semi-supervised classification

Yongquan Shi, Yueyang Pi, Zhanghui Liu, Hong Zhao, Shiping Wang

Abstract: Highlights•Propose a GCN-based framework for multi-view semi-supervised classification.•Analyze the node propagation process in GCNs from an information-theoretic perspective.•Alleviate the over-smoothing problem without sacrificing the model’s expressive power.•Stabilize both the forward inference and backward propagation processes in GCNs.