Support Vector Machine with Improved Particle Swarm Optimization Model for Intrusion Detection

Intrusion Detection System (IDS) is a computer-based data system which purports to observe attacks against computer systems and networks or, against any information system. Its job is to supervise the utilization of such system to detect any insecure states. IDS detect attempts and active misuse of the scheme either by lawful users of the information systems or by outside parties to abuse privileges or exploit security vulnerabilities. It gets information about target system to perform diagnosis on security status. Data mining techniques used for intrusion detection are classification, clustering, frequent pattern mining and mining data streams. The classification method called Support Vector Machine (SVM) has been used to provide potential results for the intrusion detection problem. Nevertheless, the practicability of SVM is affected due to the trouble of selecting appropriate SVM parameters and feature selection. The optimization algorithm, Particle Swarm Optimization (PSO) is applied to pick out the optimized parameters for the SVM. For initializing the population of PSO, Optimal Latin Hypercube Design (OLHD) is applied. The OLHD follows the space-filling property of attributes in the design space. The proposed OLCPSO-SVM model is employed to an intrusion detection problem in the KDD Cup 99 data set. The experimental results show that the OLCPSOSVM method can reach a higher detection rate than regular SVM algorithms.