A-DOGE
Fast Attributed Graph Embeddings via Density of States
A-DOGE is an unsupervised graph embedding algorithm which:
- is scalable
- is task-agnostic
- works independently on each graph
- can capture local and global graph properties
- can capture node labels and attributes
- can capture information from arbitrary bands of graph spectra
Saurabh Sawlani, Lingxiao Zhao, Leman Akoglu. Fast Attributed Graph Embeddings via Density of States. In 21st IEEE International Conference on Data mining (ICDM). 2021.