This was L2S at NeurIPS 2025

The DFG Research Unit Learning to Sense represented by Shashank Agnihotri, Marius Bock, Mishal Fatima, Vaishnavi Gandikota, Margret Keuper, Jovita Lukasik, and Michael Möller had organized a workshop about the joint optimization of imaging pipelines and neural networks at the Neural Information Processing (NeurIPS) Conference in San Diego on Dec 7th 2025.

The workshop featured several highly influential keynote speakers in the field including Michael Felsberg (Linköping University), Michael Brown (York University and Samsung AI), Daisuke Iso (Senior Researcher at Sony AI), Hyung-Sin Kim (Seoul National University), and the L2S Mercator Fellow Felix Heide (Princeton University). The latter four joined a panel discussion on the importance of dedicated data acquisition (see foto), about which Hyung-Sin Kim also presented a position paper at the NeurIPS main conference (https://openreview.net/forum?id=xnNHXepQ9h).

The L2S workshop featured a poster session including 13 posters. Suhani Grover received the workshop’s best poster award for her presentation on “Hankel Dynamic Mode Decomposition for Radar-Based Respiratory Sensing and Tracking”. Details about the workshop including all extended abstracts and papers can be found at https://sites.google.com/view/l2s-workshop/home.