Several students have won important awards, such as Frederik's PhD thesis being awarded the best Nordic thesis in image analysis and computer vision at the SCIA conference; Stas's paper on identifiability through geometry was accepted at ICML 2025; Andrea's work on contactomorphisms for stable control was accepted at ICML; Miguel's new preprint proposed a method to apply the sketched Lanczos uncertainty score to generative models; two AISTATS papers covered efficient sampling of highly correlated approximate posteriors and pullback metrics in GP-LVMs; Hadi's preprint extended the NCDS paper; Marco and Hrittik published a new preprint on constructing approximate posteriors that are guaranteed not to underfit.
Research Experience
Gave lectures at multiple international conferences such as MTNS, ELLIS Summer School, etc.; supervised students in completing their PhD theses; participated in various research projects.
Background
I do machine learning and computer vision research, where I work with geometric models of observed data. The research tends to follow two directions: use geometric constructions to design models, or model constraints on data using geometry.
Miscellany
Discussed AI with high school students at Experimentarium; lectured at the ScaDS.ai summer school which took place in Leipzig Zoo; gave a lecture at the Niels Bohr Institute.