Statistics Seminar - 03/20/25

Abstract:

The task of detecting weak signals is undertaken pervasively in practice, including analyzing effects of items in questionnaires, omics biomarkers and nutrients. Often such discovery requires large sample sizes, which can be very costly. Homogeneity pursuit provides a novel solution to such challenge in that weak signals are grouped to form signal-sets with increased statistical power.  This approach requires essentially to perform a simultaneous operation of grouping and estimation in data analyses.  Recently we developed a new paradigm of supervised homogeneity pursuit via mixed integer optimization, which provides a conceptually simple and computationally straightforward machinery with the use of suitable constraints in optimization.  Our proposed simultaneous operation is also expanded with a new development of uncertainty quantification by means of adversarial noise perturbation. Our new toolbox will be illustrated by several real-world data examples. 

About the Speaker:

Dr. Song is Professor of Biostatistics at the School of Public Health in the University of Michigan, Ann Arbor. He received his PhD in Statistics from the University of British Columbia, Vancouver, Canada in 1996. He has published over 230 peer-reviewed papers and graduated 26 PhD students and trained 6 postdoc research fellows. Dr. Song’s current research interests include data integration, distributed inference, high-dimensional data analysis, longitudinal data analysis, mediation analysis, spatiotemporal modeling, and applications in medicine and public health. He collaborates extensively with researchers from nutritional sciences, environmental health sciences, chronic diseases, infectious disease, aging and nephrology. He is IMS Fellow, ASA Fellow and Elected Member of the International Statistical Institute. Dr. Song now serves as Editor of the Annals of Applied Statistics and Associate Editor of the Journal of American Statistical Association and the Journal of Multivariate Analysis.