L2-traditional methods
Contents
Traditional Methods in Graph Learning
node-level features
- degree
- Centrality measures: degree centrality, betweenness centrality, eigenvector centrality, closeness centrality, clustering coefficent
- 进一步引申GDV
edge-level features
我觉得这个很像我们的vae encoder想法
距离表征容易理解……
邻接矩阵的幂（类似于归纳法 略）
graph-level features
核方法（但是skip） bag of sth…
recall when you need
