Jakob Nylöf

Hi, welcome to my page!

I’m a doctoral student at the Automatic Control Laboratory and Risk Analytics and Optimization Chair at EPFL, as well as the Automatic Control Laboratory at ETH Zürich. I am jointly supervised by Prof. Giancarlo Ferrari Trecate, Prof. Daniel Kuhn and Prof. John Lygeros. My research interests lie in the intersection of control theory, stochastic programming and robust optimization, with a key focus on distributionally robust optimal control.

Biography

I received a BSc in Engineering Physics from KTH Royal Institute of Technology in 2021 and a MSc in Mathematics in 2024 from the same institution. My master thesis was on deep Q-learning in continuous time, supervised by Prof. Boualem Djehiche. During my studies, I was a summer researcher at KTH under the supervision of Prof. Karl H. Johansson and at the University of Michigan where I was supervised by Prof. Necmiye Ozay. I also had an exchange semester at ETH Zürich.

Distributionally Robust Optimization

In Distributionally Robust Optimization (DRO), the goal is to minimize a risk measure, such as the expected value of a loss function, while accounting for uncertainty in the underlying probability distribution. Instead of assuming a known distribution, DRO minimizes the worst-case expectation over a set of plausible distributions (the ambiguity set). This ensures robustness by finding the best decision under the worst possible distribution, safeguarding against model uncertainty.