Genetic Algorithms as a Tool tor Feature Selection in Machine Learning

This paper describes an approach being uplored to improve the usefulness 0/ machine leanaing techniques lor generating classijication rules/or complex. real world dt:ua. The approach involves the use 0/genetic algorithms as a irON end" to traditional rule induction systems in order to ideNify and select the best .subset o//eatures to be used by the rule induction system. This approach has been implemeNed and rested on difficult tature classification problems. The results are encouraging and indicate signijicanJ advantages to the preseNed approach in this domain.