B0372
Title: Discrete smoothing kernels
Authors: Marianthi Markatou - University at Buffalo (United States) [presenting]
Abstract: Kernels are essential elements in the construction of learning systems and have received considerable attention in the machine learning literature. In statistics, kernels are used as tools for achieving specific data analytic goals such as density estimation or goodness of fit testing. We consider the problem of creating smoothing kernels for multinomial and product multinomial models. Our construction is based on the properties of continuous time Markov chains. We will discuss an algorithm for the construction of these kernels and exemplify its use in smoothing ordered categorical data. Furthermore, using these constructions, we will discuss independence tests that are analogues of the chi-squared test of independence.
