Guest lecture: Contributions to Machine Learning by Means of Genetic Programming
Dr. Sébastian Mahler from Université dy Littoral Côte d'Opale, Calais, France, will give a guest lecture on Tuesday, December 9th at 10 o'clock in lecture room B222.
Abstract:
In this talk, we focus on how to use a particular heuristic, namely Genetic Programming, to solve Machine Learning applications. We start with a brief introduction to Genetic Programming (GP) (Koza, 1992) that produces and adapts (optimizes) a population of programs like in Darwin's Evolution theory: the main principals of theory help to solve a given problem. This technique is related to Genetic Algorithms (Holland, 1975) (also called Evolutionary Algorithms). Then we see a concise view on some problematics and results of my work. The last part of the talk explains how to apply the meta-heuristic of boosting (Freund and Shapire 1995) with GP. In this context, we show explicit motivations to use of a particular learning criteria: maximizing the Area Under the ROC (Receiver Operating Characteristics) Curve. We also present some results with our approach (ROCBoost) that has been used with two evolutionary heuristics.