Research at DSF

Research at our institute rests on two pillars that support each other: mathematical theory development and experimental work in a robotics lab. Both are integrated within information geometry — a framework that treats learning and inference as geometric processes. One can think here, for example, of a distance measure not in physical space, but between a learning system and its goal. Of course, this is only an example. What matters is that theory guides the experiment, and the experiment, in turn, reveals where the theory must be refined. It is not about passing on knowledge, it is about discovering it.

Our research seminar takes place on Thursdays at 11:00 in Room 5.002, HIP One. The talks are public, held on site and via Zoom, and feature external guests alongside institute members. No registration is required.

Since 2021, we have organised the Mathematics of Machine Learning (MoML) conference series. The series began at ZiF, Bielefeld University, and continues at TU Hamburg.

Our publications list covers all years and is updated continuously. It includes journal articles, conference papers and preprints resulting from our work in information geometry and embodied intelligence.

Doctoral and habilitation theses mark the completion of a research phase. Each thesis contributes a piece to the larger vision — a puzzle piece within a context, rather than an end in itself.

In the published interview, Nihat Ay discusses the foundations of the institute's work in an essayistic and personal register. The conversation touches on the two-pillar concept, the role of mathematics and the limits of imagination.

Our contributions to science communication include the exhibition *Der Körper denkt mit*, shown on the MS Wissenschaft vessel and at the Deutsches Museum Bonn. The exhibit demonstrates how body design can take over computation, allowing the brain to become smaller.

For questions about research collaboration or doctoral positions, please contact dsf-office@tuhh.de.