sMGC: A Complex-Valued Graph Convolutional Network via Magnetic Laplacian for Directed Graphs.
Jie ZhangBo HuiPo-Wei HarnMin-Te SunWei-Shinn KuPublished in: CoRR (2021)
Keyphrases
- directed graph
- complex valued
- real valued
- random walk
- convolutional network
- graph structure
- recurrent neural networks
- convolutional neural networks
- undirected graph
- directed acyclic graph
- strongly connected
- multi valued
- linear equations
- activation function
- graph structures
- graph mining
- feed forward
- edge detection
- multiscale
- web pages
- coarse to fine
- disjoint paths
- training set