My research interests include software security and software structure, also called architecture.
In particular, I am interested in security guarantees for intelligent algorithms
with a focus on, but not limited to, neural networks. The property of such algorithms to withstand
malicious attacks or to avoid unintended and harmful behavior has been previously defined as robustness.
In other settings, it has been defined as safety or assurance.
I am currently working on robustness of machine learning from an algorithm perspective, but also
from a system and systems-of-systems perspective. Regarding the former, I am interested in reducing
the resources needed for training robust models.
Regarding the latter, I am interested in the impact such algorithms have on software structure and (traditional)
software engineering because (1) some security concerns regarding the deployment of intelligent algorithms
can be mitigated through software architecture and (2) late trends in
end-to-end learning have shaken the way software is designed and promise to replace a substantial
number of software components.
The research is funded by NWO through the i-CAVE project, where the focus is on autonomous driving.