Sensible Functional Classification

Abstract

A summary of my research on functional/longitudinal data analysis since 2011 will be given in the talk. The topics include multidimensional functional principal component analysis, a functional approach to deconvolve dynamic neuroimaging data, inverse regression for longitudinal data and sensible functional classification. The major focus of the talk is on our recently developed classification approach, where we extended Fisher’s linear discriminant analysis to both functional data and longitudinal data for general c-category classification problems. In the end of the talk, I will briefly introduce some of my ongoing and future research projects.