Cyber-Threat-Detection

The Aegean WiFi Intrusion/threat Dataset (AWID2) was prepared by George Mason University (U.S.) and University of the Aegean (Greece) to study cyber threat detection. Specifically this project is focused on the classification using ML algorithms of impersonation attacks. Impersonation attacks can be particularly serious as they can cause unauthorized or malicious users to access the network. The datasets used for this project is a modified version of the initial AWID2 dataset where just impersonation attacks and normal traffic are considered. Both the test and the train sets are balanced (having the number of attacks being equal to the normal traffic). A substantial part of this experiment is to try out different feature generarion techniques (such as Autoencoders) to generate new features. Different feature selection techniques in combination with a variety of ML classification models are trained on the original and generated data (such as Random Forest, SVC, MLP) and evaluated to determined the most suitable ML model for the impersonation attacks problem.