Keywords
Neural Network, Explainable AI, NIDS

Technologies
Keras, SHAP, SLIPS, Streamlit

Abstract

This project investigates the use of Deep Neural Networks to classify malicious network traffic and improve the detection of suspicious behaviour within modern networks.

The system is integrated into an open-source Intrusion Detection System, allowing network traffic to be analysed and classified in a more intelligent and automated way. To make the results easier to understand, a frontend dashboard is used to display the model’s predictions in a clear and accessible format.

In addition, the project applies the SHAP explainable AI framework to provide insight into how the model reaches its decisions. This increases transparency, helps users understand why traffic is flagged as suspicious, and makes the overall system more trustworthy for practical cybersecurity use.