Low-Rank Approaches to Graphon Learning in Networks

The graphon is a powerful framework for modeling large-scale networks, but its estimation remains a significant challenge. In this paper, we propose a novel approach that directly leverages a low-rank representation of the graphon for parsimonious modeling. This representation naturally yields both a low-rank connection probability matrix and a low-rank graphon -- two tasks that are often infeasible in existing literature -- while also addressing the well-known identification issues in graphon estimation. By exploiting the additive structure of this representation, we develop an efficient sequential learning algorithm that estimates the low-rank connection matrix using subgraph counts and reconstructs the graphon function through interpolation. We establish the consistency of the proposed method and demonstrate its computational efficiency and estimation accuracy through extensive simulation studies.