Entropy of Exchangeable Random Graphs

arXiv2302.01856MaRDI QIDQ6508680FDOQ6508680

Abstract: Model complexity remains a key feature of any proposed data generating mechanism. Measures of complexity can be extended to complex patterns such as signals in time and graphs. In this paper, we are concerned with the well-studied class of exchangeable graphs. Exchangeability for graphs implies a distributional invariance under node permutation and is a suitable default model that can widely be used for network data. For this well-studied class of graphs, we make a choice to quantify model complexity based on the (Shannon) entropy, resulting in graphon entropy. We estimate the entropy of the generating mechanism of a given graph, instead of choosing a specific graph descriptor suitable only for one graph generating mechanism. In this manner, we naturally consider the global properties of a graph and capture its important graph-theoretic and topological properties. Under an increasingly complex set of generating mechanisms, we propose a set of estimators of graphon entropy as measures of complexity for real-world graphs. We determine the large--sample properties of such estimators and discuss their usage for characterizing evolving real-world graphs.

This page was built for publication: Entropy of Exchangeable Random Graphs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6508680)