I’m currently working on my Master’s final project, and I’m brainstorming different ways to get started on this. My current plan is to use three VMs: one for the AI attacker, one for the Metasploitable who will be attacked and one to train the attacker. I can’t seem to come up with an idea on what kind of neural network I will be using to train the attacker. I have very limited experience with artificial intelligence in general, and none when applying it to hacking. If anyone has any ideas as to what kind of neural networks can work in a situation like this or any experience using neural networks in hacking, then I would love to listen to your ideas, suggestions and stories!
Just a disclaimer, I am not looking for any solutions or methodologies to get my project started. I am simply looking for ideas based on theory and experience. Thank you for your time!
EDIT: I should add that the testing of this project will be conducted within an isolated network to avoid any spills and leaks.