Feature Selection Based on Genetic Algorithm and SupportVector Machine for Intrusion Detection System

One of the most common problems in existing detection techniques is the high curse of dimensionality, due to multidimensional features of the network attack data. This paper investigates the performances of genetic algorithm (GA) with support vector machine (SVM) classification method for feature selection, the forward feature selection algorithm (FFSA) and linear correlation feature selection (LCFS) in detecting different types of network attacks. In particular, the feature selection capability of GA, FFSA and LCFS has been studied. In this work GA, FFSA and LCFS have been implemented and tested on KDD CUP 1999 dataset. The results have shown that all of the algorithms are capable of achieving about 99% detection rate at different number of reduced features. GA with SVMand LCFS require only 21 features, while FFSA requires 31 features to detect the attacks effectively. In addition, the false positive results shown by all of the algorithms are comparatively low, between 0.43% and 0.59% when the detection rate is almost perfect.