zbMATH — the first resource for mathematics

Dependence modeling rule mining using multi-objective genetic algorithms. (English) Zbl 1157.68412
Gammerman, A. (ed.), Artificial intelligence and applications. Machine learning. As part of the 26th IASTED international multi-conference on applied informatics. Calgary: International Association of Science and Technology for Development (IASTED); Anaheim, CA: Acta Press (ISBN 978-0-88986-710-9/CD-ROM). 278-283 (2008).
Summary: This work investigates the use of multi-objective genetic algorithms in the mining of accurate and interesting rules for the dependence modeling task. Dependence modeling is a generalization of the classification task in which a set of goal attributes is used. A multi-objective evolutionary environment named MO-miner was implemented based on the family of algorithms called non-dominated sorting genetic algorithms. Two desirable properties of the rules being mined - accuracy and interestingness - are simultaneously manipulated. MO-miner keeps the metrics related to these properties separated during the evolution, as different objectives used in the fitness calculus in a Pareto-based approach. The environment was applied to a public domain database named Nursery. The results obtained by MO-miner had been compared with those generated by a standard GA in order to identify the benefits related to the multi-objective approach.
For the entire collection see [Zbl 1154.68012].
68T05 Learning and adaptive systems in artificial intelligence
68W05 Nonnumerical algorithms
PDF BibTeX Cite