Common Randomized Shortest Paths (C-RSP): A Simple Yet Effective Framework for Multi-view Graph Embedding

Abstract

Real-world data sets often provide several types of information about the same set of entities, showing us how they interact from different viewpoints. These data sets are well represented by multi-view graphs, which consist of multiple edge sets across the same set of nodes. Combining multiple views improves the quality of inferences drawn from the underlying data, which has led to increased interest in developing efficient multi-view graph embedding methods. We propose an algorithm, C-RSP, that generates a common (C) embedding of a multi-view graph using Randomized Shortest Paths (RSP). This algorithm generates a dissimilarity measure between nodes by minimizing the expected cost of random walks between any two nodes across all views of the graph, in doing so encoding both the local and global structure of the graph. We test C-RSP on both real and synthetic data and show that it outperforms benchmark algorithms at embedding and clustering tasks.