Benedikt Brückner

My PhD project focuses on verification of AI systems, in particular neural networks. Nowadays, a broad variety of systems employ AI algorithms to perform certain tasks. However, these models do not lend themselves to traditional verification methods, and for some systems, it may not even be possible to understand their internal logic at all. Since so-called “adversarial attacks” have proven that most systems have significant faults and can easily be fooled by handcrafting appropriate inputs, there is a dire need for methods that can formally check the correctness of these systems. Existing algorithms to provide such guarantees often suffer from scalability issues, making it impossible to verify large state-of-the-art neural networks. I hope that I can make contributions in these fields during my time at the CDT. 

I chose to do my PhD with the CDT in STAI because of the benefits that the CDT offers to its students. One can acquire broad knowledge in different areas related to Safe and Trusted AI through the lectures and seminars that are offered. Having a cohort of students interested in similar topics is a great opportunity to meet new people and support each other during the PhD. The research project that I work on provides me with the opportunity to combine my previous knowledge in optimisation and machine learning to engage in state-of-the-art research, so it seemed to be a perfect fit for me. 


Undergraduate Qualification: BSc in Industrial Engineering and Management, Karlsruhe Institute of Technology 

Masters Qualification: MSc in Industrial Engineering and Management, Karlsruhe Institute of Technology 

Research Experience: During my studies, I worked as a research assistant supporting research in the field of robust and stochastic optimisation.