Feature selection using fuzzy objective functions

One of the most important stages in data preprocessing for data mining is feature selection. Real-world data analysis, data mining, classification and modeling problems usually involve a large number of candidate inputs or features. Less relevant or highly correlated features decrease, in general, the classification accuracy, and enlarge the complexity of the classifier. Feature selection is a multicriteria optimization problem, with contradictory objectives, which are difficult to properly describe by conventional cost functions. The use of fuzzy decision making may improve the performance of this type of systems, since it allows an easier and transparent description of the different criteria used in the feature selection process. In previous work an ant colony optimization algorithm for feature selection was presented, which minimizes two objectives: number of features and classification error. Two pheromone matrices and two different heuristics are used for each objective. In this paper, a fuzzy objective function is proposed to cope with the difficulty of weighting the different criteria involved in the optimization algorithm.