College of LAS Events

Back to Listing

If you will need disability-related accommodations in order to participate, please email the contact person for the event.
Early requests are strongly encouraged to allow sufficient time to meet your access needs.

Statistics Seminar - Bing Li (Penn State University)

Event Type
Seminar/Symposium
Sponsor
Department of Statistics
Virtual
wifi event
Date
Dec 1, 2022   3:30 - 5:00 pm  
Views
125
Originating Calendar
Department of Statistics Event Calendar

Title: On sufficient graphical models
by Bing Li and Kyongwon Kim

Abstract:

We introduce a Sufficient  Graphical Model by applying the recently developed nonlinear sufficient dimension reduction techniques to the evaluation of conditional independence. The graphical model is nonparametric in nature, as  it does not  make distributional assumptions  such as the Gaussian or copula Gaussian assumptions. However, unlike a fully nonparametric graphical model, which relies on the high-dimensional kernel to characterize conditional independence,    our graphical model is based on conditional independence given a set of sufficient predictors with a substantially reduced dimension. In this way we avoid  the  the curse of dimensionality that comes with a high-dimensional kernel. We develop the population-level properties,  convergence rate, and variable selection consistency of our estimate. By  simulation comparisons and an analysis of the DREAM 4 Challenge data set, we demonstrate that our method outperforms the existing methods when the Gaussian or copula Gaussian assumptions are violated, and its performance remains excellent in the high-dimensional setting.

link for robots only