Classification with application to Functional Data based on Gaussian process

Abstract

In this paper, we briefly introduce four methods for functional classification. To compare the effects of the four models, we generate the data from Gaussian process based on a functional mixed-effects model, square exponential kernel is used in random-effect term to describe the nonlinear structure of the data. The outcomes show that the two functional classification models have a better prediction correct rate than the two machine learning classification models.