This repository includes the code to the paper "Increasing Detection Rate for Imbalanced Malicious Traffic using Generative Adversarial Networks" of EICC 2024 (Note: The "DIAX" model here is called "combined".). It includes 4 models: A GAN, WGAN, XGBoost and the DIAX model.

First run

conda env create -f environment.yml

to get all needed dependencies for the python code.

Please also download the cicids2018 data set, this needs to be put in path: /mnt/md0/files/cicids2018/ or change the path in prepocess.py.

Run in src/ :

    python3 main.py 1 1 1

Explanations of the numbers are given in the main.py file. Please note that the preprocess.py file has a FEATURES_DROPPED list that can be commented in if the reduced model should be tested. Then also add the appropriate just use the combined model with frags in the main.py