HIN2Grid: A disentangled CNN-based framework for heterogeneous network learning

作者：

Highlights：

• We propose a disentangled CNN-based model for fast heterogeneous graph learning.

• HIN2Grid avoids suffering complex and recursive neighborhood expansion in GCNs.

• HIN2Grid is better than GCN-based models in computation and memory saving.

• Dual attention mechanisms improve the accuracy and robustness of HIN2Grid.

• Extensive experiments demonstrate the superiority of HIN2Grid in various tasks.

摘要

•We propose a disentangled CNN-based model for fast heterogeneous graph learning.•HIN2Grid avoids suffering complex and recursive neighborhood expansion in GCNs.•HIN2Grid is better than GCN-based models in computation and memory saving.•Dual attention mechanisms improve the accuracy and robustness of HIN2Grid.•Extensive experiments demonstrate the superiority of HIN2Grid in various tasks.

论文关键词：Network embedding,Heterogeneous graph learning,Graph convolutional network,Data mining

论文评审过程：Received 19 February 2021, Revised 20 May 2021, Accepted 27 August 2021, Available online 11 September 2021, Version of Record 24 September 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115823