On 1
October 2026, the spin-off project Forma Metis started at the Scientific
Computing Center (SCC). It is funded by the Helmholtz Association through
Helmholtz Enterprise, the spin-off programme of its Initiative and Networking
Fund, which supports researchers at Helmholtz centres in turning research
results into companies.
Forma Metis
develops software for interpretable machine learning. Many AI models are black
boxes: they predict well, but no one can say how. The Forma Metis software
instead produces models that are compact mathematical formulas. Domain experts
can read them, check them and compare them with their own knowledge. The models
are transparent by design, not explained after the fact.
Readable
models are useful wherever predictions must be understood and justified. In
research and development, for example in materials science and pharma, the
formula itself shows the pattern behind the data. In regulated decisions, for
example in finance and health care, the models can be checked and audited.
The founders, Dr. Luigi Sbailò and Dr. Lucas Foppa, work on the project at SCC. Prof. Luca Ghiringhelli, head of the research group Computational and Data Science in Materials Research (CDSMR), is the project's scientific advisor.
Further information is available on the project website.
Contact at SCC: Dr. Luigi Sbailò