Graph label prediction based on local structure characteristics representation

作者：

Highlights：

• Centrality regions are more representative than random partial substructures.

• Divide the selected paths into similar set and dissimilar set to increases model generalization.

• Design a new index to measure the similarity between two paths.

• Vote for multi-layer MLP prediction results to increase accuracy.

摘要

•Centrality regions are more representative than random partial substructures.•Divide the selected paths into similar set and dissimilar set to increases model generalization.•Design a new index to measure the similarity between two paths.•Vote for multi-layer MLP prediction results to increase accuracy.

论文关键词：Graph classification,Graph neural network,Betweenness centrality node,Feature fusion,Characteristics representation

论文评审过程：Received 28 March 2021, Revised 25 September 2021, Accepted 7 January 2022, Available online 10 January 2022, Version of Record 19 January 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108525