Intrusion Detection using Machine Learning - Final Exam Project

Authors: Piotr Drapiński, Ivana Jasna Caltagirone, Arianna Sammarchi, Nikolaj Viktor Brøchner Pettersson

This repository contains the technical part of the final project for the course "Cybersecurity Foundations and Analytics." The project focuses on building a machine learning-based intrusion detection system to predict network intrusions using the UNSW-NB15 dataset that can be found here.

Project Overview

The project focuses on detecting network intrusions using the UNSW-NB15 dataset, which includes:

• Attack Types: DoS, worms, Backdoors, Fuzzers, and more.
• Labels: Binary classification of attack (1) and normal (0).

Methodology

The technical analysis includes:

• Data preprocessing and feature engineering.
• Exploratory data analysis (EDA).
• Model training and evaluation using Decision Tree, Random Forest, Logistic Regression, XGBoost, and ensemble methods.
• Hyperparameter tuning for model optimization.
• Evaluation metrics: Accuracy, Precision, Recall, F1-Score, AUC-ROC.

Conclusion

This project demonstrates the application of machine learning in cybersecurity, highlighting the efficiency of ml-based systems in that space. The technical analysis provides insights into key predictors of network intrusions, aiding in proactive security measures.