Dynamic graph convolutional networks

作者：

Highlights：

• First neural network approaches to classify dynamic graph-structured data.

• We propose two novel techniques: WD-GCN and CD-GCN.

• These techniques are based on combination of graph convolutional units and LSTM.

• Semi-supervised classification of sequence of vertices.

• Supervised classification of sequence of graphs.

摘要

•First neural network approaches to classify dynamic graph-structured data.•We propose two novel techniques: WD-GCN and CD-GCN.•These techniques are based on combination of graph convolutional units and LSTM.•Semi-supervised classification of sequence of vertices.•Supervised classification of sequence of graphs.

论文关键词：

论文评审过程：Received 14 May 2018, Revised 21 May 2019, Accepted 13 August 2019, Available online 13 August 2019, Version of Record 22 August 2019.

论文官网地址：https://doi.org/10.1016/j.patcog.2019.107000