Anomaly intrusions detection based on support vector machines with bat algorithm

Intrusion Detection Systems(IDS) have become an essential part of every security framework. These systems rely on monitoring and detection of intrusions, thus composing an additional line of defense. Several paradigms have been applied for implementing IDS. In this paper we propose a NIDS model based on Information Gain for feature selection and Support Vector Machines(SVM) for the detection component. SVM is a feed forward neural network with many advantages that comply with the requirements of an IDS. One drawback of this classification algorithm is that its performance depends on some input parameters. Our solution for this optimization problem is to apply swarm intelligence. We test our approach with the NSL-KDD data set and show that our model can obtain better results than regular SVM or PSO-SVM.