Vitality based feature selection for intrusion detection

Intrusion detection system is the process to monitor network traffic to detect possible attacks. In recent time, network traffic increasing rapidly. There are plenty of research today focused on feature selection or reduction, as some of the features are irrelevant and degrade the performance of an intrusion detection system. By eliminating some of features, we can improve the performance of classification algorithm. In this paper, we evaluate the performance of feature selection methods, such as Correlation Based Feature Selection (CFS), Information Gain (IG), Gain Ratio (GR), Feature Vitality Based Reduction Method (FVBRM). We propose a modification to FVBRM by changing the parameter True Positives Rate (TPR) into False Positives Rate (FPR) and by applying Naïve Bayes classifier on reduced dataset to measure the result of our feature selection method. The results of modified FVBRM indicate that selected attributes provide better performance for intrusion detection system.