Presenter: Aditya Ranade, Graduate Student, Department of Statistics, Iowa State University

Title: Relevance Determination Of Discretized Functional Inputs In Gaussian Process Regression

Abstract: In the context of Gaussian process regression, functional input and scalar output is a common scenario. Treating the functional input into discretized vector profiles allows the use of GP under the time varying input framework. Finding the appropriate distance between the vectorized profiles becomes key to the performance of the GP. We look at different distance functions using the ordered input profiles and explore weight (relevance) functions from the Asymmetric Laplace Functional weight (ALF) family.