Feature selection using multi-objective evolutionary algorithms : application to cardiac SPECT diagnosis

摘要

An optimization methodology based on the use of Multi-Objective EvolutionaryAlgorithms (MOEA) in order to deal with problems of feature selection in datamining was proposed. For that purpose a Support Vector Machines (SVM) classifierwas adopted. The aim being to select the best features and optimize the classifierparameters simultaneously while minimizing the number of features necessaryand maximize the accuracy of the classifier and/or minimize the errors obtained.The validity of the methodology proposed was tested in a problem of cardiac SingleProton Emission Computed Tomography (SPECT). The results obtained allowone to conclude that MOEA is an efficient feature selection approach and the bestresults were obtained when the accuracy, the errors and the classifiers parametersare optimized simultaneously.