Title: Mesoscale two-sample testing for networks

with Peter W. MacDonald, University of Waterloo

Tuesday, December 3rd
Votey Hall, Room 209
3:00 PM - 4:00 PM

Abstract: Networks arise naturally in many scientific fields as a representation of pairwise connections. Statistical network analysis has most often considered a single large network, but it is common in a number of applications, for example, neuroimaging, to observe multiple networks on a shared node set. When these networks are grouped by case-control status or another categorical covariate, the classical statistical question of two-sample comparison arises. In this work, we address the problem of testing for statistically significant differences in a given arbitrary subset of connections. This general framework allows an analyst to focus on a single node, a specific region of interest, or compare whole networks. Our ability to conduct mesoscale testing on a meaningful group of edges is particularly relevant for applications such as neuroimaging and distinguishes our approach from prior work, which tends to focus either on a single node or the whole network. Our approach can leverage all available network information and learn informative projections that improve testing power when a low-dimensional latent network structure is present.

Bio: Dr. Peter W. MacDonald is an Assistant Professor in Statistics & Actuarial Science at the University of Waterloo, specializing in statistical analysis of multiple and dynamic networks. Previously a postdoctoral scholar at McGill University (2023-2024), he completed his PhD at the University of Michigan under Drs. Elizaveta Levina and Ji Zhu. His research focuses on network analysis methods, with notable publications in Biometrika and the Journal of Machine Learning Research. Dr. MacDonald holds both an MMath and BMath from Waterloo and has received several honors, including an NSERC Postdoctoral Fellowship and Michigan's Outstanding Dissertation Award.