Statistical Discriminant Analysis in Finite and Infinite Dimensional Spaces

Subhajit Dutta

Abstract

In this talk, we shall discuss some methods of supervised classification of finite dimensional multivariate data and infinite dimensional sequence data with a common and finite state space. For finite dimensional multivariate data, we will discuss the maximum depth classifier based on projection depth and related ideas. We also consider some classification problems under l_p-symmetry of the distribution of data and discuss classifiers based on l_p depth. In the context of sequence data, we will discuss some classifiers based on Markov models. We will also consider some non-parametric classifiers based on distributions of k-words.