Wolfgang Stammer (PostDoc)

I am a Postdoctoral Researcher in the Computer Vision and Machine Learning Group at the Max Planck Institute for Informatics, supervised by Prof. Bernt Schiele, and an associated member of the Research Training Group "Neuroexplicit Models" at Saarland University. I completed my PhD with distinction at the AI & ML Lab, TU Darmstadt, under the supervision of Prof. Kristian Kersting.

I am interested in Compositional Reasoning in AI. Particularly, compositionality as a driving principle for AI models that are understandable, controllable, and generalizable. A central question in my work is: what is the right unit of composition for a given reasoning task, whether that be a concept, an object, a relation, a symbolic program, or a logical proposition, and how should those units be combined to support transparent and robust reasoning? This connects three interrelated challenges: discovering compositional representations with minimal supervision, grounding them in visual and linguistic inputs, and reasoning faithfully over them. Rather than treating interpretability as a trade-off against expressiveness, I see compositionality as the mechanism that can bridge the two. For more infos see my, visit my personal webpage.