Analysis of Network Intrusion Detection and Potential Botnets Identification Using Selected Machine Learning Techniques

Patryk Zabawa, Michal Kedziora

Abstract: The paper presented here is centered around the analysis of network attack detection using machine learning. It starts by examining the development and categorization of network attacks, and then provides an overview of conventional detection methods. Additionally, it conducts a thorough analysis of the ninth scenario in the CTU-13 dataset and outlines the steps involved in preparing for the experiment using this dataset and conducting feature engineering. The results section primarily focuses on the Random Forest and Decision Tree algorithms, as well as the classification of botnets using a Multilayer Perceptron (MLP), comparing the machine learning approaches with traditional methods of network attack detection.