Simultaneous Feature and Model Selection for High-Dimensional Data

Abstract

The paper proposes an Evolutionary-based method to improve the prediction performance of Support Vector Machines classifiers applied to both artificial and real-world datasets which suffer from the curse of dimensionality. This method performs a simultaneous feature and model selection to discover the subset of features and the SVM parameters' values which provide a low prediction error. Moreover, it does not require a pre-processing step to filter the features so it can be applied to a whole dataset.