Header Workshop on Foundation models

Helmholtz-ELLIS Workshop on Foundation Models in Science

The Helmholtz-ELLIS Foundation Models in Science workshop is an extraordinary gathering of leading minds at the intersection of artificial intelligence and scientific research. The carefully curated lineup of speakers represents a remarkable cross-section of cutting-edge AI research and its transformative potential across multiple scientific domains.

The first day focuses on foundational large language models (LLMs) and multimodal AI, featuring speakers who are literally shaping the future of machine learning. Their talks promise to explore the latest advancements in AI model development, from resource efficiency to multimodal capabilities.

The second day is particularly exciting, showcasing AI's revolutionary applications in specific scientific fields, such as materials science, astronomy, and physicsEach speaker represents an institution at the forefront of integrating AI into their respective domainsdemonstrating how AI is not just a technological tool, but a fundamental research methodology transforming scientific inquiry.

With speakers from Meta, Microsoft, ETH Zurich, Simons Foundation, ELLIS, and other prestigious institutions, this workshop represents a pivotal moment in interdisciplinary scientific research, showcasing how artificial intelligence is becoming an indispensable partner in understanding and solving complex scientific challenges.
The full line-up of speakers will be announced soon on this website.

If you are interested, kindly complete your application by clicking on the Register button below. The deadline for registration is 14 February 2025 at 23:59 CET. After approval of your application you will be asked to pay the registration fee, which is:

  • MSc and PhD students: 150€
  • Other participants: 250€

The fee coverslogistic costs, coffee breaks, lunch and the conference dinner. Travel and accommodation are self-organized. The workshop dinner on March 18 will take place at the Natural History Museum.

Confirmed speakers

Tian Xie, Principal Research Scientist, MSR (Microsoft Research AI Science), Cambridge, U.K.

Anna Scaife, Professor of Radio Astronomy, University of Manchester, Manchster, U.K.

Piotr Bojanowski, Research Scientist, Meta (Facebook AI Research), Paris, France

Gaël Varoquaux, Research Director, INRIA (National Institute for Research in Digital Science and Technology), Paris, France

Michal Valko, CMO Stealth Startup & Inria & ENS MVA; Paris, France

Matthias Bethge, Computational Neuroscience & Machine Learning, Universität Tübingen, Tübingen, Germany

Shirley Ho, Senior Research Scientist, Cosmology, Foundation Models for Science, CCA, Flatiron Institute, Flatiron Institute - Princeton University, New York, USA

Sarath Chandar, Associate Professor, Department of Computer and Software Engineering / MILA – the Quebec AI Institute, Ecole Polytechnique de Montréal, Montréal, Quebec, Canada

Günter Klambauer, Associate Professor for “Artificial Intelligence in Life Sciences” at the LIT AI Lab and the Institute for Machine Learning of the Johannes Kepler University, Linz, ELLIS Austria, Vienna/Linz, Austria

Kangwook Lee, Assistant Professor at the Electrical and Computer Engineering Department and the Computer Sciences Department, University of Wisconsin-Madison, Wisconsin, USA

Lena Maier-Hain, Professor at Heidelberg University, and Head of the division Intelligent Medical Systems (IMSY) at the German Research Cancer Center (DKGZ), Heidelberg, Germany

Willie Neiswanger, Assistant Professor of Computer Science, University of Southern California, USA

Lucas Beyer, OpenAI, Zürich, Switzerland

Tim Hempel, Microsoft Research AI for Science, Berlin, Germany

Michael Gastegger, Microsoft Research AI for Science, Berlin, Germany

Johannes von Oswald, ETH & Google Zürich, Switzerland

Directions to Dinner

Venue

Alte Münze
Molkenmarkt 2
10179 Berlin
Germany

Time

-

Program

Day 1

TimeSpeakerTitle
08:15Registration 
08:45Dagmar Kainmueller (HFMI / MDC)Welcome Day 1
 Session 1 - Chair: Dagmar Kainmueller (HFMI / MDC)
09:00Gaël Varoquaux (Director Inria)Tabular Foundation Models: Priors for Numbers and Strings
09:45Michal Valko (CMO Stealth Startup & Inria & ENS MVA)Gamification of Large Language Models 
10:30Coffee break 
11:00Piotr Bojanowski (Meta FAIR)Applications of Large-scale Self-supervised Representation Learning
11:45Shirley Ho (Simons Foundation & NYU)Building Pan-scientific Foundation Models: Polymathic AI
12:30

Lunch

Group photo at 13:15

 
 Session 2 - Chair: Jilles Vreeken (CISPA)
13:30Kangwook Lee (U. Wisconsin-Madison)Bridging Natural Language Understanding and Classical Machine Learning: From LIFT to LLM-Lasso for Predictive Modeling
14:15Johannes von Oswald (ETH & Google Zurich)The Current Status on Transformer Alternatives for Language Modelling
15:00Coffee break 
15:30Lena Maier-Hein (DKFZ & U. Heidelberg)Rethinking Foundation Model Evaluation 
16:15Matthias Bethge (U. Tübingen)Cosmopolitan Foundation Models
 Panel Discussion 
17:00

Moderator: Stefan Kesselheim

Panel: Michal Valko, Shirley Ho, Kangwook Lee, Johannes von Oswald

Open LLMs in AI4Science
18:00Otmar D. Wiestler (Helmholtz Association)Reception and Welcome by the President of the Helmholtz Association
20:00 Dinner at the Natural History Museum

Day 2

TimeSpeakerTitle
08:55Fabian Isensee (HFMI / DKFZ) Welcome Day 2
 Session 3 - Chair: Marie Piraud (Helmholtz-AI / HM)
09:00Tim Hempel & Michael Gastegger (Microsoft Research AI for Science)Scalable Emulation of Protein Equilibrium Ensembles with Generative Deep Learning
09:45Tian Xie (Microsoft Research AI for Science)Accelerating Materials Design with AI Emulators and Generators 
10:30Coffee break 
11:00Anna Scaife (U. Manchester)Foundational Representations for Astrophysics
11:45Lucas Beyer (OpenAI)Training Vision-Language Foundation Models
12:30Lunch 
 Session 4 - Chair: Fabian Isensee (HFMI / DKFZ)
13:30Sarath Chandar (MILA & Polytechnique Montreal)Efficient Foundation Models for Proteins and Molecules
14:15Willie Neiswanger (USC)METAGENE-1: A Metagenomic Foundation Model for Pandemic Monitoring
15:00Coffee break 
15:30Günter Klambauer (Johannes Kepler University & ELLIS Linz)xLSTM-based foundation models for Life Sciences
 Panel Discussion 
16:15

Moderator: Martin Schultz

Panel: Michael Gastegger, Tian Xe, Anna Scaife, Sarath Chandar

AI4Science - Opportunities and Challenges 
17:15 End of Workshop

Organizers

Helmholtz Foundation Model Initiative – Synergy Unit 
&
European Laboratory for Learning and Intelligent Systems


Supported by