Konstantin Kirchheim

I am a PhD student in computer science at Otto von Guericke University Magdeburg. My research focuses on safe and trustworthy machine learning under distribution shift, particularly out-of-distribution detection and neuro-symbolic reasoning.

I study how learned representations and structured knowledge can be combined to detect inputs and situations in which machine-learning models may behave unreliably. My work has appeared at venues including ICML, WACV, and CVPR Workshops.

I was previously a visiting graduate student at the University of Waterloo. I also maintain pytorch-ood, an open-source Python library for out-of-distribution detection with deep neural networks.

Outside my main research, I occasionally work on data-mining and self-hosting projects. Some of these have developed into independent websites, including SWORM and extra-mining.

Google Scholar · GitHub · DBLP · ORCID