Detection Coverage of ML-Based Classification Models for Network Intrusion Detection by the Application of Varied Pre-Processing Techniques

In order to prevent network vulnerabilities and data exploitation in real-time, it is crucial to implement appropriate cyber security solutions. With a comprehensive strategy, mission-critical systems can be safeguarded from intrusion by an effective intrusion detection system. In order to identify network breaches, this research presents a comprehensive ML-based security solution. The method employs an ensemble feature selection methodology and an ensemble supervised ML framework. We also compare various feature selection and ML model approaches. Improving the accuracy and reducing the false positive rates (FPR) of generic detection systems is the main goal of this study. In this experiment, NSL-KDD, UNSW-NB15, and CICIDS2017 are the datasets that were utilised. With a false alarm rate of only 0.5% and an intrusion detection rate of 99.3%, our detection model outperforms previous methods.