ECAI-2002 Conference Paper

[full paper]

AREX – Classification Rules Extracting Algorithm Based on Automatic Programming

Vili Podgorelec, Peter Kokol, Ivan Rozman

The paper presents a hybrid classification method of BNF grammar-based genetic programming and evolutionary decision tree induction, customized for the rule induction according to a layered hierarchical scheme – the AREX approach. It incorporates two original, independent evolutionary algorithms which together solve the problem of automatic classification rules induction. The method is applied to five real world databases (from medicine and software engineering) and the results are compared to those obtained with C5/See5 to evaluate the method’s efficiency. Ideally, this paper will inspire future research in this same area and along similar lines.

Keywords: Data Mining and Knowledge Discovery, Machine Learning, Genetic Algorithms

Citation: Vili Podgorelec, Peter Kokol, Ivan Rozman: AREX – Classification Rules Extracting Algorithm Based on Automatic Programming. In F. van Harmelen (ed.): ECAI2002, Proceedings of the 15th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2002, pp.330-334.