01.10.2026
Berlin, October 1, 2026. The Berlin Initiative for Applied Foundation Model Research has had RamanBench accepted at the 40th Conference on Neural Information Processing Systems (NeurIPS 2026), one of the world's most important conferences for artificial intelligence and machine learning. Mario Koddenbrock will present the results in December 2026 in Paris.
"Our first goal with RamanBench was to make the few public datasets accessible. They are scattered and exist in very heterogeneous forms on all kinds of servers. We only gradually realized how large the research gap actually is here. There is a lot of great methodological work in this field, but it is usually evaluated on just a single dataset. Comparability between different studies is rarely given. The ML community is already far ahead on this front. There are standard benchmarks and protocols, without which you can barely publish a new paper anymore. We tried to push this practice forward for Raman spectroscopy. At the same time, we are offering the ML community an exciting new task. Large labs and startups have already reached out to us.
"But our contribution is not only about making data accessible and developing protocols. Together with TU Berlin and many national and international partners, we released 17 new datasets. Since the preprint came out, we have already received further datasets. Our hope is that RamanBench will become a 'living benchmark', with more datasets and models joining over time.
"And I think that the DFG project and the doctoral center are creating a really great collaboration between the Berlin universities of applied sciences and beyond.
"Right now, though, we are just looking forward to Paris. I think having a NeurIPS paper is the dream of every ML PhD student. I am glad to have the opportunity to do this kind of top-level research here," says Mario Koddenbrock. He and Dr. Christoph Lange (TU Berlin) are joint first authors of the study.
Project partners
- Berliner Hochschule für Technik (BHT)
- Hochschule für Technik und Wirtschaft Berlin (HTW Berlin)
- Technische Universität Berlin (TU Berlin), Department of Bioprocess Engineering
- Einstein Center Digital Future, Berlin
- Hochschule Niederrhein, Krefeld
- VTT Technical Research Centre of Finland, Oulu
- KWS SAAT SE & Co. KGaA, Einbeck
