Graph Clustering Using One-Bit Comparison Data

Graph clustering with partial adjacency information has garnered significant interest in data processing and learning applications. In this paper, we propose an extension of a set of non-convex programs, generally referred to as Comparison-Aided 2-Hard and Comparison-Aided 2-Soft Clustering programs, that rely on matrix factorization formulations to process one-bit comparison data. Numerical results confirm not only the usefulness of the proposed methodologies, but also their potential to be used in future data processing applications. To the best of our knowledge, this is the first work in the area of graph clustering that relies only on comparison information to accomplish the clustering task.