Abstract :

Cybersecurity experts frequently want assistance from an automated procedure that filters and sorts network attacks. Before implementing specific preventative measures to safeguard networks, you must identify the type of attack. Some have proposed building Network Intrusion Detection (NID) systems on top of various Machine Learning (ML) models. However, a variety of circumstances influence their effectiveness. An ML model built on an unequal dataset, for example, can favor attack types that are too prevalent. However, the ML model may not do as well in majority classes if you only consider how well it performs in minority classes. We offer a Network Intrusion Detection (NID) system that use Convolutional Neural Networks (CNN) to classify various types of assaults and address the issue of unbalanced datasets. The suggested system's performance is contrasted with that of current systems that employ various data balancing techniques, including Synthetic Minority Oversampling