Big Ten Research Talk: Yingjie Li

Date: Time: 3:00 pm–4:00 pm
Hardin Hall Room: 49 North Wing Basement
Additional Info: HARH
Contact: Stacey Herceg, 472-2903, sherceg2@unl.edu
Yingjie Li, a visiting graduate student from Michigan State University, will discuss “High Dimensional Discriminant Analysis for Spatially Correlated Data” in a 3 p.m. Nov. 8 lecture in the Hardin Hall Auditorium.

The talk is free and open to the public.

Linear discriminant analysis is one of the most classical and popular classification techniques. However, it performs poorly in high-dimensional classification. Many sparse discriminant methods have been proposed to make LDA applicable in high dimensional case. One drawback of those methods is the structure of the covariance among features is ignored.

Through research, Li proposes a new procedure for high dimensional discriminant analysis for spatially correlated data. Penalized maximum likelihood estimation is developed for feature selection and parameter estimation. Tapering technique is applied to reduce computation load.

The theory shows that the method proposed can achieve consistent parameter estimation, features selection, and asymptotically optimal misclassification rate. Extensive simulation study shows a significant improvement in classification performance under spatial dependence.

This research is motivated by how to using voxel level brain imaging data for classification. It can be applied for classification to other data sets with spatial correlation among features, such as image data from biology, agriculture, natural resource and medical study.

Additional Public Info:

Download this event to my calendar