Detect Cyber Intrusion Using Machine Learning

A cyber attack happens every 39 seconds. An intrusion detection system acts as the first line of defense to detect these attacks. In this project, we’ll implement machine learning based classifiers that can accurately detect and classify several types of cyber attacks. The classifiers will learn patterns of benign and malicious activities from existing network traffic datasets. Using this learning, the classifiers will detect and flag malicious intrusions.

In this project, we’ll use SIMARGL2021, a publicly available dataset that contains benign and malicious network traffic. Firstly, we’ll explore the dataset to understand its basics, such as the number of features, type of attacks, and redundancy in the data. Next, we’ll visualize the data to understand the different labels and their proportion in the datasets. Then, we’ll train and test machine learning models using multiple classifiers such as random forest, decision tree, and Gaussian Naive Bayes. Finally, we’ll assess the accuracy of the trained classifiers.