ABSTRACT / Most empirical studies of complex networks do not return direct, error-free measurements of network structure. Instead, they typically rely on indirect measurements that are often error-prone and unreliable. A fundamental problem in applied network science is how to make the best possible estimates of network structure given such unreliable data. In this talk, I will present a Bayesian framework for reconstructing networks from empirical data in any format, even when the data contain substantial measurement error and when the nature and magnitude of that error is unknown. The framework allows for estimation even when the network measurements are reflective of multiple possible structures---like proximity data measured both on weekdays (network 1) and weekends (network 2). As a result, the framework solves multiple statistical problems at once: clustering measurements, identifying underlying network(s), and estimating the parameter of the measurement models. I will demonstrate how to implement estimation in this framework with powerful computational methods such as Hamiltonian Monte Carlo and Gibbs sampling, and give several examples from the applied network science literature.
BIO / Jean-Gabriel Young is an Assistant Professor of Mathematics and Statistics at The University of Vermont, VT, USA. He also is a faculty of the Translational Global Infectious Diseases Research Center (TGIR) and the Vermont Complex Systems Center (VCSC). His research is at the intersection of statistical inference, epidemiology, and complex systems. Before joining UVM, he was a James S. McDonnell Foundation Fellow at the Center for the Study of Complex Systems of the University of Michigan, mentored by Prof. Mark Newman. He obtained his Ph.D. in Physics from Université Laval, under the guidance of Prof. Louis J. Dubé and Prof. Patrick Desrosiers.