Coco Bögel
About
I am a PhD student in the Machine Learning Interpretability subgroup under supervision of Giuseppe Casalicchio and Bernd Bischl since January 2025. My current focus is on functional decompositions (like the fANOVA) and regional or tree-like interpretability methods (like MOB or GADGET).
More generally, my interests lie in better understanding AI and machine learning systems and in the safety of these systems. This includes topics like explainability, robustness, generalization and distributional shift.
Before my PhD, I studied Mathematics at TU Berlin, focusing on mathematical theory of AI and ML and researching deep learning for particle simulations in my Master's thesis. I also gathered some experience in software development with C++ during an internship at think-cell.
Contact
You can write me an E-Mail at any time. It may sometimes take some time for me to answer, but I will answer it in any case. You can also find me on the LMU Mattermost Server (@coco_boegel).
Institut für Statistik
Ludwig-Maximilians-Universität München
Ludwigstraße 33
D-80539 München
Room 344, 3rd floor
c.boegel [at] stat.uni-muenchen.de
Research Interests
- Model-agnostic machine learning interpretability
- Functional decompositions
- Regional feature effect methods
- Robustness of machine learning systems (e.g. concerning distributional shift)
- AI safety and AI alignment