ELLIS: Enabling Excellence in AI Research in Europe

The Institute of Science and Technology Austria (ISTA) is home to the ELLIS Unit Vienna, a locally organized unit of the European Laboratory for Learning and Intelligent Systems (ELLIS). As a member of this continent-wide artificial intelligence (AI) research excellence network, the unit distinguishes itself by its broad spectrum and interdisciplinarity. Embedded into the ISTA ecosystem, the unit focuses on excellence in advancing machine learning (ML) foundations and applications in science.

Today’s world is shaped by the rapid advancements in artificial intelligence (AI) and machine learning (ML) that we have witnessed in recent years. But these technologies surrounding us have capitalized on the interplay between the academic and industrial spheres. In this respect, European academic researchers at the forefront of the field understood early on that their continent offered a unique advantage: the possibility to shape AI developments in the public sphere. This idea fueled a European-wide grassroots movement that would become the European Laboratory for Learning and Intelligent Systems (ELLIS), a synonym for excellence in AI and ML research in Europe.

Academic freedom in AI advancement

By aiming to push the frontiers of European AI research and ML applications in science, this grassroots movement of European researchers set the cornerstone for ELLIS in 2018. Institute of Science and Technology Austria (ISTA) professor Christoph Lampert, an ML and computer vision expert, took up the torch to shape the ELLIS Unit Vienna at ISTA as unit director. This was among the initial group of units to take shape within a year. “Together with peers throughout Europe, we set a high priority on developing AI research on our continent. We knew that AI research in Europe would play a key role in balancing industrial interests,” says Lampert. Thus, the first ELLIS units set out to help research in the public sphere shape the narrative of AI advancements. “With ELLIS, we seek to emphasize academic freedom in AI advancement. We are persuaded that scientific excellence flourishes in curiosity-driven research environments,” Lampert adds.

Five years of enabling excellence at ISTA

Since its start with six ISTA professors in 2019, the ELLIS Unit Vienna has grown to include eight core faculty members and several collaborators. The research groups of Dan Alistarh, Alex Bronstein, Monika Henzinger, Vladimir Kolmogorov, Christoph Lampert, Francesco Locatello, Marco Mondelli, and Matthew Robinson explore the theory and foundation of ML, ML applications to accelerate scientific progress in different scientific fields, and advance ML in technology. The unit’s core faculty members have independently secured multiple ERC grants, of which six have been active since the unit took shape in 2019, as well as industry funding.

The ELLIS Unit Vienna distinguishes itself by its broad spectrum, reflecting ISTA’s diversity and interdisciplinarity. Its researchers work on four complementary axes: machine learning algorithms and theory, optimization, applications in computer vision and language modeling, and other applications in various scientific disciplines.

Rallying talents and inspiring collaboration

One of ELLIS’s strengths is its central call for promising PhD students, explains Francesco Locatello, deputy director of the ELLIS Unit Vienna. Coupled with the focus on mobility—a central aspect of the ELLIS PhD and postdoc program—ELLIS grants young talents access to a network of reputed AI researchers, events, and workshops across Europe. By requiring that ELLIS graduate students have a “primary host” at their home institution and an “exchange host” at another unit, ELLIS fosters synergies between multiple institutions in Europe. “The prestige aspect of ELLIS gives us a competitive advantage in hiring graduate students in AI and ML,” says Locatello.

A good example of the ELLIS spirit is PhD student Riccardo Caddei’s project “Smoke and mirrors in causal downstream tasks” (arXiv preprint). This work explores a crucial issue in causal inference—the potential for biases to creep into downstream tasks, even when the underlying causal model is well understood. This collaboration included Locatello as Caddei’s primary host at ISTA, exchange host Cordelia Schmid at the National Institute for Research in Digital Science and Technology (Inria), France, and the group of social immunity researcher Sylvia Cremer at ISTA.

At ISTA, multiple activities of the ELLIS Unit bring people on campus together to learn about the latest advancements in AI and ML. The activities include on-campus networking opportunities, the ELLIS Unit Lunch, and regular ELLIS talks by renowned international researchers. In addition, ELLIS PhD students at ISTA present their research at local and international meetings. This year, the International Conference on Machine Learning (ICML 2024) was organized in Vienna. Thus, the ELLIS Unit Vienna seized the opportunity to couple this international meeting with an ELLIS event at ISTA, the Future of Machine Learning Symposium. This symposium brought together top minds from academia and industry to ISTA to explore the latest progress and emerging trends in the field.

Accelerated computing, a Cluster of Excellence, and a tech transfer ecosystem

Beyond enabling excellence in academic AI research, the ELLIS Unit Vienna is also well embedded in ISTA’s ecosystem, which includes xista, the hub for science-based innovation and tech transfer on campus. This year’s innovation exchange conference at ISTA, bigX, was themed “The Future of AI”. Being part of this ecosystem fosters the connection between the ELLIS Unit Vienna and various ML companies and startups in and around Vienna. Members of the ELLIS Unit Vienna also have access to cutting-edge computational infrastructure, such as an international standard NVIDIA GPU cluster for generative AI and machine learning. This resource is mainly used by the unit’s faculty while also enabling applications in the life sciences, physics, and computer science. Also this year, four ISTA professors, including three ELLIS Unit core faculty members, became key researchers in the Cluster of Excellence “Bilateral AI”, opening up further opportunities for networking and collaborations in Austria. This Cluster of Excellence across six Austrian institutes and universities has been endowed with 33 million euros over five years, 60 percent of which are contributed by the Austrian Science Fund (FWF).

“ELLIS has set the foundation for a European-wide AI network to enable progress, tech innovations, and positive societal impacts of research in machine learning. This unique network increases Europe’s sovereignty in AI research. At ISTA, multiple synergies make our location especially attractive for potential PhD applicants and postdocs who want to be part of our journey and help shape the future of AI and ML research,” concludes Locatello.