Title: Domain selection for Gaussian processes
Authors: Nicolas Hernandez - UCL (United Kingdom) [presenting]
Gabriel Martos - Fundacion Universidad Torcuato Di Tella (Argentina)
Abstract: A novel domain selection methodology is proposed for high-dimensional Gaussian processes. We use the Kullback-Leibler divergence to introduce a divergence curve as a tool to select the domain of the largest divergence between two processes. The proposed method learns and infers about the subinterval of the domain that better discriminates the classes of functions. Throughout a Monte Carlo experiment, we show the accuracy of the proposed method in the estimation of the true domain of largest divergence.