Information Theory in Data Mining
Presenters: Matthijs van Leeuwen, Katholieke Universiteit Leuven, Belgium; Arno Siebes, Universiteit Utrecht, Netherlands; Jilles Vreeken, Max-Planck Institute for Informatics, Germany and Saarland University, Germany
Tutorial Website: http://usefulpatterns.org/itdm/
Abstract
In the last decade Information Theoretic methods slowly but surely became popular in the data mining community for selecting the best model for the data at hand. In this tutorial we present an overview of these methods. Starting from the basics, to how one defines and finds good solutions using Information Theory, to how such models provide highly competitive solutions to many data mining tasks.