MFGCN: Multi-faceted spatial and temporal specific graph convolutional network for traffic-flow forecasting

Jingwen Tian, Liangzhe Han, Mao Chen, Yi Xu, Zhuo Chen, Tongyu Zhu, Leilei Sun, Weifeng Lv

Abstract: Highlights•We propose Multi-Faceted Spatial Temporal Graph Convolutional Network(MFGCN).•We capture dynamic temporal relations for different nodes in node-specific module.•We capture dynamic spatial relations across time in time-specific mudule.•An adaptive pattern sharing mechanism is designed to share patterns across layers.•MFGCN outperforms state-of-the-art methods on four real-world datasets.