This repository contains the implementation of a Machine Learning-based Intrusion Detection System (IDS) utilizing tree-based algorithms. The project demonstrates the use of classification techniques to identify and prevent potential security threats in network systems.
- Application of machine learning to detect intrusions in network traffic.
- Evaluation of tree-based algorithms such as:
- Decision Tree
- Random Forest
- Extra Trees
- Comprehensive preprocessing and feature engineering steps.
- Detailed analysis of the model's performance using metrics like accuracy, precision, recall, and F1-score.
- Insights into improving the IDS with advanced machine learning techniques.
The project uses the CIC-IDS 2018 dataset, which provides a detailed representation of real-world network traffic, including normal and malicious activities. This dataset is widely used in IDS research for benchmarking and evaluating model performance.
- Includes a variety of attacks such as DoS, DDoS, brute force, botnet, and more.
- Simulated with realistic traffic distribution over multiple days.
- Rich set of features including protocol types, packet sizes, flow durations, and more.
The dataset can be downloaded from the official Canadian Institute for Cybersecurity website.
- Python 3.8 or above
- Miniconda/Anaconda for managing dependencies
- Libraries:
scikit-learnpandasnumpymatplotlibimblearn