Recently, nonparametric approaches for modeling relational data based on exchangeable graph models have been gaining increasing interest. Relational data are typically encoded in the form of arrays, and the development of exchangeable graph models relies on a generalization of de Finetti’s theorem to exchangeable arrays due to Aldous and Hoover. The key object underlying such models is referred to as a graphon, a notion introduced by Lovász and Szegedy as limits of graph sequences. This talk will attempt to survey some recent literature on the theory of exchangeable random graphs and the estimation of graphons, drawing connections to applications in network analysis such as link prediction, community detection, and network comparison. The goal of this talk is to initiate discussions and collaborations on this relatively new topic in the Purdue statistics community.

Associated reading:

Bickel, P. J. and Chen, A. (2009). A Nonparametric View of Network Models and Newman–Girvan and Other Modularities. Proc. Natl. Acad. Sci. USA 106:21068-21073.

P. Orbanz and D. M. Roy (2015). Bayesian Models of Graphs, Arrays and Other Exchangeable Random Structures. IEEE Trans. Pattern Anal. Mach. Intell. 37(2):437-461.

Gao, C., Lu, Y. and Zhou, H. H. (2015). Rate-Optimal Graphon Estimation. Ann. Statist. 43(6):2624–2652.

Seminar Information

  • Speaker: Jiasen Yang
  • Date: Thursday, March 31, 2016
  • Time: 4:30 pm - 5:20 pm
  • Location: Math Library Lounge

Slides

Slides can be downloaded here.