AI Based Network Intrusion Detection System

The growth of using digital technology has created new challenges in keeping the information secure and safe. Attacks such as Denial of Service, Probe, Remote to Local, User to Root and other malicious activities can affect systems and data. Old existing security systems identify threats by comparing with the previously known pattern. However these methods may not identify the new or unfamiliar threat. In this project, we developed an AI-Based Network Intrusion Detection System using Machine Learning and Deep Learning. We used network intrusion datasets such as NSL-KDD, UNSW-NB15, CICIDS2017 and IoT-23.Before the process of analysis the data will be checked , organised and will be converted into suitable form for processing. Different techniques are being applied to address differences in the number of examples available for each category. An Autoencoder is being used to obtain an clear representation of the input data. A Transformerbased model is used for classification, while CNN and DNN models are also used for comparison. Our system will support both the binary classification and multiclass classification. Binary classification will identify whether the traffic is normal or an attack, and the multiclass classification identifies the type of attack. We also used SHAP to explain the model predictions and understand which features influenced the result. A webbased dashboard is provided to display the predictions and model results. The main objective of our project is to develop a complete AI-based intrusion detection system that can process the data more effectively and identify different variety of attacks and provide results that will be clear and understandable.