16 Jun 2025
A European consortium is developing robust AI systems that can deal with a huge variety of data sources. The chair of Professor Björn Ommer at LMU, where the Stable Diffusion foundation model was developed, is participating in the project.
Professor Björn Ommer | © Ansgar Pudenz
Through the ELLIOT (European Large Open Multi-Modal Foundation Models For Robust Generalization On Arbitrary Data Streams) project, the European Commission is funding the development of a new generation of general-purpose AI models that are built for real-life, data-rich applications.
At the heart of ELLIOT is the development of multimodal foundation models: AI systems that learn general knowledge and patterns from massive amounts of data of various types — from videos, images, and text to sensor signals, industrial time series, and satellite feeds — and transfer the knowledge learned in a task to a wide variety of other tasks. Unlike current foundation models which face significant challenges in terms of generalization capabilities and support for multimodal data, ELLIOT’s models will be capable of robust generalization across dynamic, noisy, and temporally-evolving multimodal data streams.
With the participation of 30 research institutions and companies from a dozen European countries, the consortium is coordinated by the Information Technologies Institute of the Centre for Research and Technology Hellas (CERTH-ITI) in Greece. It includes computer scientist Professor Björn Ommer, who leads the Computer Vision & Learning Group at LMU. Aside from Ommer’s team, other German partners include researchers from Tübingen, Jülich, and Saarbrücken. The program is being funded to the tune of 25 million euros for a period of four years through the Horizon Europe initiative.