M.A. Thesis:

Combination of Support vector machine with particle optimization algorithm for intrusion detection system

In this paper, the framework of the intrusion detection by the use of a new method is proposed. The combination of the binary and connected particles swarm optimization (PSO) algorithm is respectively and synchronously used for the feature selection and the parameters optimization of the support vector machine. The support vector machine is a powerful algorithm for data classification, which its function depends on a variety of parameters such as violation factor c and kernel parameter ?. The selection of an appropriate kernel function can also improve the detection rate. Moreover, the selection of the useful feature among a variety of features in data set not only increases SVM function but also decreases time and intricacy in calculation. Therefore, it’s an optimization issue which can be solved by the exploratory algorithms. In this article, the multi-class support vector machine is used with the cross validation k-fold method as a classification system. The function of the proposed method has been evaluated by testing the NSL_KDD data set, and the empirical results demonstrate that the proposed method in comparison with other methods operates better in detection and false alarm rate.

Keywords: intrusion detection,particle swarm optimization,support vector machine,cross validation, Feature selection