Graph convolutional neural networks with node transition probability-based message passing and DropNode regularization
作者：
Highlights：
• A new message passing formulation for graph convolutional neural networks is proposed.
• An effective regularization technique to address over-fitting and over-smoothing.
• The proposed regularization can be applied to different graph neural network models.
• Semi-supervised and fully supervised learning settings are considered.
• The proposed method is evaluated via extensive experiments on benchmark datasets.
摘要
•A new message passing formulation for graph convolutional neural networks is proposed.•An effective regularization technique to address over-fitting and over-smoothing.•The proposed regularization can be applied to different graph neural network models.•Semi-supervised and fully supervised learning settings are considered.•The proposed method is evaluated via extensive experiments on benchmark datasets.
论文关键词：Graph convolutional neural networks,Graph classification,Node classification,Geometric deep learning
论文评审过程：Received 8 June 2020, Revised 27 December 2020, Accepted 10 February 2021, Available online 18 February 2021, Version of Record 4 March 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.114711
