A Survey on Feature Ranking by Means of Evolutionary Computation

Abstract

The paper presents a review of current evolutionary algorithms for feature rankingin data mining tasks involving automated learning. This issue is highly important as real-worldproblems commonly suer from the curse of dimensionality. By weighting the signicanceof each attribute from a data set, the less inuential indicators can be disposed of beforelearning actually takes place, making the task become easier and less noisy. Moreover, forseveral practical domains of application, such as medicine for instance, a ranking of the mostindicative attributes for a diagnosis are as vital as the computational learning support for thenal decision taking. Evolutionary algorithms are one of the most frequently used heuristicsfor a diverse range of tasks, due to their simple and exible nature. Therefore, the currentstudy investigates the numerous recent trends in employing the evolutionary computation eldfor the subject of feature ranking.