The Fifth Jyotiprasad Medhi Memorial Annual Lecture was delivered by Prof. Shankar Bhamidi, Department of Statistics and Operations Research, University of North Carolina, Chapel Hill on September 27 at 5:30 PM (Online).

Title of the Lecture: What does math probability have to say about Network science ?

Venue: Online Mode

Abstract:

Research into the applications of networks has exploded in a myriad of domains ranging from applications in areas such as neuroscience, sociology and political science, urban planning etc and has stimulated intense effort in the development of methodological tools in areas such as mathematics, statistics, computer science and statistical physics. The aim of this talk is to describe “in words” four recent findings that to my mind emphasize the importance of probability as a technical tool for answering such questions:

Seed detection: Imagine a network evolving from a starting seed graph (e.g. “patient zero”) and growing over time (e.g. epidemic on a network). If we just had topological (connectivity) information of the final state of the network with no information on time ordering, can one estimate the initial seed?

For epidemics on network models such as the configuration model, what determines the (positive probability) of emergence of an epidemic (that infects a non-trivial fraction of the population) vs infections dying out quickly?

For models of dynamic networks that undergo a shock resulting in a change point, what are potential issues in detecting such a change point even with observation of all time slices of the network? Can one understand MCMC algorithms for simulating models of networks such as exponential random graphs? What insight can this provide on the relevance of such models to empirical data?