Improved Relevance Vector Machine (IRVM) classifier for Intrusion Detection System

Intrusion detection is the most significant research area in online applications to avoid intrusion activities. The foremost goal of the present research is to use the Relevance Vector Machine which can recognize extract intrusion activities involved in the Intrusion Detection System. Classification and feature selection are implemented by Improved Relevance Vector Machine and Gaussian Firefly Algorithm, respectively. The proposed work contains three phases such as preprocessing, feature selection and classification, and it would increase the classification accuracy. Preprocessing uses the technique Kalman filtering which focuses on missing values in the given Knowledge discovery in databases. Gaussian Firefly Algorithm selects the most relevant and optimal features, thereby increasing overall execution speed. Then Improved Relevance Vector Machine identifies intrusion attacks efficiently by extracting more relevant vectors and thus classifying maximum likelihood values. The experimental result concludes that Improved Relevance Vector Machine algorithm provides greater performance in terms of precision, recall, specificity and accuracy.