Tech Talk: Sufficient Dimension Reduction with Simultaneous Variable Selection for Ultrahigh Dimensional Data

Location: CCHS or web
Date: Thu: 09/06/18 12:00 PM - 1:00 PM

"Sufficient Dimension Reduction with Simultaneous Variable Selection for Ultrahigh Dimensional Data”

Hi there, let Wei Qian feed your mind September 6, 2018.

Sufficient dimension reduction (SDR) is known to be a powerful tool for achieving data reduction and data visualization in regression and classification problems. In this work, we study ultrahigh-dimensional SDR problems and propose solutions under a unified minimum discrepancy approach with regularization. When p grows exponentially with n, consistency results in both central subspace estimation and variable selection are established simultaneously for important SDR methods. Special sparse structures of large predictor covariance are also considered for potentially better performance. The proposed approach is equipped with new algorithms to efficiently solve the regularized non-convex objective functions and a new data-driven procedure to determine structural dimension and tuning parameters. We further study high dimensional SDR approach for censored data in survival analysis. Real data examples are presented to demonstrate the promise of our proposal in genomics and cancer studies.

 For more information and to RSVP, click here. 



Location:
CCHS or web
Date:
Thu: 09/06/18 12:00 PM - 1:00 PM
Continuing Ed Credits: 1

Contact:
Deena Chisholm
phone:
302-733-5868