150 participants from 90 different institutions and with 40 different nationalities: The ELLIS Winter School on Foundation Models organized by the ELLIS Unit Amsterdam was a vibrant gathering place for young scientists and top AI researchers from all over Europe. During three days, the school provided an in-depth look into how Europe is guiding its own research agenda in the crucial field of foundation models and contributed to bringing the European research community closer together.
Foundation models - their origin, analysis and development - have typically been associated with the US and Big Tech. Yet, a critical share of important insights and novel approaches do come from Europe, both within academia and industry. Recognizing this fact, the ELLIS Winter School in Amsterdam gathered some of Europe’s most renowned researchers in this field who presented their work to an audience of 150 PhD students.
Among the speakers of this ELLIS flagship event were ELLIS Fellows such as AI pioneer Sepp Hochreiter from the Johannes Kepler University Linz (‘xLSTM: Memories for large language models’), Max Welling from the University of Amsterdam (‘ML to design free energies and free energies as a foundation for ML’), Zeynep Akata from the Helmholtz Munich and the Technical University of Munich (TUM) (‘Interpretable visual and language assistants (VLAs)’) and Raquel Fernández from the University of Amsterdam (Large language models and human cognition), as well as ELLIS member and ELLIS PhD Award winner Mathilde Caron from Google Research (‘How to train computer vision systems with less manual supervision’) and ELLIS members Hilde Kühne from the University of Bonn and MIT-IBM Watson AI Lab (‘Video and language’) and Christian Rupprecht from the University of Oxford (‘Learning from and with large models’).
Further contributions came from industry researchers such as Isran Misra from Meta AI (‘Improving generative models for vision: high quality videos and precise image control’), Thomas Wolf from Hugging Face (‘A little guide to building large language models in 2024’), Jean-Baptiste Alayrac from Google DeepMind (‘Large scale multimodal models from Flamingo to Gemini’), and Tim Salimans from Google Research (‘Generating images and video with diffusion models’).
Poster sessions and social activities completed the program
Not only did the PhD students have the opportunity to exchange directly with top AI experts at the event, but they also got the chance to present their own research during poster sessions, discuss their results, and explore possible collaboration with peers.
The participants also had the opportunity to explore scenic Amsterdam on a canal cruise trip while getting to know each other and chatting over different topics in a more relaxed setting.
“The ELLIS Winter School on Foundation Models provided 150 students from 90 institutions from all over Europe with the latest advances on this revolutionary AI topic. The lectures by renowned experts from academia and industry covered the many facets of foundation models and were not only intellectually pleasing, but also provided the students with hands-on advice and long-term research perspectives. Overall, it was a most rewarding experience to host this school in Amsterdam”, says Prof. Dr. Cees Snoek, Director of the ELLIS Unit Amsterdam and organizer of the ELLIS Winter School on Foundation Models.
“The ELLIS Winter School in Amsterdam was a fantastic opportunity to explore foundation models with leading European researchers. I was particularly excited to see how geometric principles from computer vision are still crucial, and the talks on best practices for training large models were incredibly valuable. It was inspiring to connect with fellow PhD students from across Europe, particularly those working on causality, representation learning, and embodied AI”, highlights Frederik Nolte, ELLIS PhD student (DPhil candidate in Engineering Science at the University of Oxford) and participant of the ELLIS Winter School on Foundation Models.
The winter school was partially funded by the ELLIS Unit Amsterdam and supported by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 951847, known as ELISE - European Network of AI Excellence Centres.
About ELLIS Winter / Summer Schools
The ELLIS network thrives on the distributed efforts of its members and units. Following a call for proposals for the organisation of ELLIS Summer/Winter Schools, a set of ELLIS units was selected to host such an event. These schools focus on deep diving into machine learning research topics, showcasing the work at that unit, and providing essential training and networking activities to the program participants. The next events in line are the ELLIS Summer School on Machine Learning for Healthcare and Biology in Manchester in June, and the ELLIS Summer School on Collaborative and Generative AI in Helsinki in July.
More information
Visit the website of the ELLIS Unit Amsterdam here.
Learn more about the ELLIS PhD Program here: https://ellis.eu/phd-postdoc
Read the statement of the ELLIS Board on “AI Foundation Models: A Roadmap for Europe”.
Watch the ELLIS PhD Program video and these short video testimonials recorded by some of our students across Europe to get an impression of what it’s like to be an ELLIS PhD student.
Social media
Follow the ELLIS Unit Amsterdam on X and LinkedIn.
Follow ELLIS on X, LinkedIn, Mastodon and Facebook, and subscribe to the monthly ELLIS newsletter.
