On the comparison of crazy particle swarm optimization and advanced binary ant colony optimization for feature selection on high-dimensional data

DNA microarray technology not only can measure the expression level of thousands of genes simultaneously in one experiment but can also identify the possibility of diagnosing a disease. Microarray data consists of thousands of variables, but the available data is very little. Conventional classification methods are not effective and efficient to handle this type of data. Support Vector Machine (SVM) is a supervised machine learning method that can be used for classification on the high-dimensional dataset. Therefore, reducing data dimensions will simplify and accelerate the classification process. Feature selection will eliminate irrelevant features so that it can improve the quality and accuracy of classification and can accelerate the learning process. Several approaches have been carried out for the feature selection process, including the feature selection with wrapper-based approach. Wrapper-based algorithm used in this research is Crazy Particle Swarm Optimization (CRAZYPSO) and Advanced Binary Ant Colony Optimization (ABACO). Both of CRAZYPSO and ABACO algorithm are inspired by the movement behavior of animal in finding food sources. This research uses k-fold cross-validation accuracy to compare the CRAZYPSO and ABACO algorithm for feature selection in the case of microarray data classification using Support Vector Machine Classification. The result shows that ABACO algorithm gives better result than CRAZYPSO algorithm with higher accuracy rate and less selected features.