Student Seminar Series - September 16, 2004
University of Minnesota
School of Statistics
College of Liberal Arts

Classification with Large Number of Input Variables

Chihche Lin


Thursday, September 16, 2004
2:00 PM, B53 Ford Hall
Minneapolis, East Bank Campus

Refreshments at 1:30 PM
300 Ford Hall

Abstract

 Pattern Recognition, a very important scientific subject, concerns the classification of objects into a finite number of categories. The objects can be signal waveforms, patients with certain medical history, or unresolved images. The inferential goal is to construct a classification rule for future data based on training data. Many pattern recognition methods have been well studied. However, due to the rapid growth of information technology, a great deal of input information for each object is available but the sample sizes tend to be relatively small. Because of this, statistical inference is often problematic. Asymptotic theories usually can not be applied. Model based methods are hampered by the lack of model diagnosis. We shall focus on binary classification problems, the so called supervised pattern recognition with two category response, with concentration on a large number of input variables, i.e. covariates.