Abstract

Abstract: Ensuring robust network security is of utmost importance in the present era. To safeguard the integrity of in-network systems, numerous architectural solutions have been suggested to prevent unauthorized access by both internal and external users. Several techniques have been created to identify harmful activity on targeted machines. In some cases, an external user may engage in malicious behavior and gain illegal access to these devices. Such behavior is classified as malicious activity or intrusion. Several machine learning and soft computing algorithms have been developed to identify activities in real-time network log audit data. The data sets KKDDCUP99 and NLSKDD are commonly used to identify intruders in benchmark data sets. This study presents a method for detecting and identifying unauthorized individuals using machine learning methods. Two distinct methodologies have been suggested, namely signature-based detection and anomaly-based detection. The experimental investigation showcases the application of Principal Component investigation (PCA) and Random Forest (RF) algorithms on different data sets. It also evaluates the performance of the system in a real-time network context.Keywords: Intrusion Detection System, Network security, Naïve Bayes, PCA, Artificial Neural Network, KDDCUP99