My research is focused on the automatic generation of attack trees for automotive on-board networks in order to facilitate automated systematic cybersecurity evaluation. I'm studying the ways in which machine learning could be applied towards this problem area in order to produce attack trees which can discern emergent and unexpected behaviour in ways that traditional formal methods of attack tree generation cannot.
In the early stages of my research I am investigating how a generative model can be taught on such nuanced and complex problem domain, and how this learning can be performed on a small dataset.