Network comparison

Gesine Reinert, University of Oxford. Part of the statistics seminars series.

Many complex systems can be represented as networks, and the problem of network comparison is becoming increasingly relevant. There are many such techniques that vary from simply comparing network summary statistics to computationally costly alignment-based approaches. The challenge remains to correctly cluster networks that are of a different size and density, yet hypothesized to be structurally similar. 

In this talk, we review existing methods for network comparison and introduce a new network comparison methodology that is aimed at identifying common organizational principles in networks. The methodology is simple and intuitive and outperforms existing methods in a variety of settings ranging from the classification of chemical compounds to tracking the evolution of networks representing the topology of the internet.

The talk is based on joint work with C. Deane, R. Gaunt, L. Ospina, and A. Wegner.

Gesine Reinert, University of Oxford