Graph-structured data are pervasive in the real-world such as social networks, molecular graphs and transaction networks.

Graphs are among the most widely-used data structures in machine learning. Their power comes from the flexibility of capturing relations (edges) of collections of entities (nodes) which arise in a ...