Learning Role-based Graph Embeddings.

Random walks are at the heart of many existing network embedding methods.However, such algorithms have many limitations that arise from the use ofrandom walks, e.g., the features resulting from these methods are unable totransfer to new nodes and graphs as they are tied to vertex identity. In thiswork, we introduce the Role2Vec framework which uses the flexible notion ofattributed random walks, and serves as a basis for generalizing existingmethods such as DeepWalk, node2vec, and many others that leverage random walks.Our proposed framework enables these methods to be more widely applicable forboth transductive and inductive learning as well as for use on graphs withattributes (if available). This is achieved by learning functions thatgeneralize to new nodes and graphs. We show that our proposed framework iseffective with an average AUC improvement of 16:55% while requiring on average853x less space than existing methods on a variety of graphs.