Classical and Bayesian Methods of Smooth Global Testing for Functional Linear Models

• Activity Number: 406
• Type: Topic Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Dan Spitzner*+
• Companies: University of Virginia
• Keywords: functional data analysis ; Bayesian hypothesis testing ; smoothness ; high dimensional inference

Abstract

A global perspective treats a functional data point as a whole atom. Though high-dimensional, smooth functional data, such as would arise as curves, images, or functional profiles, reside in a severely constrained region of the parameter space, which suggests that careful tailoring of inferential procedures can avoid the hopeless loss of discriminatory power suffered by generic high-dimensional procedures. This talk will discuss approaches to incorporating a smoothness assumption into global hypothesis testing procedures for functional linear models, emphasizing an interplay between classical and Bayesian approaches, and seeking optimal configurations under asymptotic criteria.