Designing an MI-PCA based Agile Intrusion Detection System

A long-standing issue with the categorization of network traffic has been brought on by redundant and pointless data characteristics. These characteristics hinder a classifier from making correct conclusions and slow down the classification process, which is especially problematic when dealing with large amounts of data. In this research, we offer a mutual information and PCA based method that chooses the best feature for classification by analytical means. Both linearly and nonlinearly dependent data characteristics may be handled by this mutual information-based feature selection technique. The characteristics chosen by our suggested feature selection technique and decision tree are used to construct an intrusion detection system (IDS), known as MIP AD. MIP AD showed the accuracy of 99% and MIPCA based feature selection technique helped other classifiers such as LR, SVM to reduce the training time by 90%.