Detecting graph similarities and graph matching
Learning a quantitative measure of the similarity among graphs is considered a key problem. Indeed, it is a critical step for network analysis and can also facilitate many ML problems, such as classification, clustering, and ranking. Many clustering algorithms, for example, use the concept of similarity for determining if an object should or should not be a member of a group.
In the graph domain, finding an effective similarity measure constitutes a crucial problem for many applications. Consider, for instance, the role of a node inside a graph. This node might be very important for spreading information across a network or guaranteeing network robustness: for example, it could be the center of a star graph or it could be a member of a clique. In this scenario, it would be very useful to have a powerful method for comparing nodes according to their roles. For example, you might be interested in searching for individuals showing similar...