Hands-On Artificial Intelligence for Cybersecurity

The IDSGAN training dataset

First of all, IDSGAN uses the NSL-KDD dataset (http://www.unb.ca/cic/datasets/nsl.html), which contains both malicious and genuine traffic samples. These samples are particularly useful for checking the performance of IDSGAN, as they are also used by common IDS.

The NSL-KDD dataset is then used as a benchmark both to verify the effectiveness of the generator component and to allow the discriminator component to return the feedback required to create the adversarial examples. Therefore, the choice of the NSL-KDD dataset is not by chance, as the traffic data samples contain both normal and malicious traffic, subdivided into four main categories, such as probing (probe), denial of service (DoS), user to root (U2R ...