Hands-On Artificial Intelligence for Cybersecurity

Introducing IDSGAN

Also, in the case of IDSGAN, the type of attack is based on a black-box strategy, in which the implementation details and configuration of the target IDS are unknown.

The underlying GAN of IDSGAN usually includes two antagonistic neural networks in which the generator component takes care of transforming the original network traffic into malicious traffic through the crafting of adversarial examples.

The discriminator component of IDSGAN, on the other hand, deals with correctly classifying the traffic, simulating the black-box detection system, thereby providing the necessary feedback to the generator component for the creation of adversarial examples.

Even in the case of IDSGAN, the adversarial examples generated using ...