Skip to main content
1920x480pxpx_Bauer_Stefan_AH_780958
TUM | © Andreas Heddergott

Interview with Prof. Stefan Bauer Navigating the Frontiers of AI Research

Stefan Bauer is an Associate Professor at TU Munich and a Senior PI at Helmholtz AI. His publications have received multiple awards, including the Best Paper Award at the 2019 International Conference on Machine Learning (ICML). As an organizer of competitions for the application and development of AI, such as drug research or real-robot-challenge.com, he advocates for the democratization and transparency of machine learning research.

Stefan Bauer is an Associate Professor at TU Munich and a Senior PI at Helmholtz AI. His publications have received multiple awards, including the Best Paper Award at the 2019 International Conference on Machine Learning (ICML). As an organizer of competitions for the application and development of AI, such as drug research or real-robot-challenge.com, he advocates for the democratization and transparency of machine learning research.

We work on all phases of experimental design and we want to automate data collection and make data collection processes significantly more efficient!
Prof. Stefan Bauer, Associate Professor at TU Munich and Senior PI at Helmholtz AI

What (research) goals are you pursuing at Helmholtz Munich?

​​I joined Helmholtz Munich on March 1, 2023 as associate professor and director at Helmholtz AI. In my research group, we work on developing algorithms that learn causal relationships from high-dimensional inputs, explain their decisions, and adapt quickly to new problems. All these requirements are key prerequisites for robust and transformative AI-based technologies with various downstream applications. Here, we have the unique opportunity to collaborate with colleagues based at Helmholtz Munich from interdisciplinary fields to jointly develop these technologies and work towards translational use cases.

The relationship between the number of storks and the birth rate is famous for the lack of a causal connection. What happens when causality is not integrated into the AI approach of Machine Learning?

When causality is not integrated into AI and machine learning, models may detect correlations without understanding the underlying cause-and-effect relationships. This can lead to inaccurate predictions, biased decisions, or false conclusions, as the model may mistake spurious correlations (like storks and birth rates) for meaningful patterns. To address this, we aim to develop novel methods for integrating causality into machine learning models.        

In the joint project "CausalNet," we aim to achieve this and have received a BMBF grant for the next three years about 2m Euros distributed across multiple project partners.

You are one of the experts of the Helmholtz Foundation Model Initiative.
What is the main goal and your contribution?

The large-scale pretraining of neural networks presents a seismic shift in AI research across various disciplines, introducing numerous fundamental challenges and opportunities. Such a paradigm shift brings forth its own set of fundamental research questions, which need to be tackled to ensure these developments are effectively leveraged across the diverse research domains as well as all data modalities in the Helmholtz Association. For this task, the Helmholtz Foundation Initiative was started.

As part of the synergy unit, we coordinate Foundation Model Efforts in Helmholtz and build Long-term and Synergistic Foundation Model Expertise.

What inspires you to come to Helmholtz Munich?
What is typical of Helmholtz Munich for you?

Helmholtz offers an incredibly exciting and interdisciplinary research environment. At Helmholtz, outstanding research groups with leading experts from various disciplines come together. From medical research to ecology, Helmholtz Munich hosts outstanding research groups. I am looking forward to starting joint research projects, gaining inspiration for new research questions, and testing and advancing algorithms in concrete applications.

What fascinates you about your research?

The opportunity to gain insights into various application areas and collaborate with different experts in an exciting environment. Machine learning is currently a highly sought-after research field with rapid cycles of innovation, and it is fascinating to see how many other fields incorporate it into their own research. Being part of this active research community is truly enjoyable. The interdisciplinary collaboration with experts and intrinsically motivated researchers is fascinating and continues to excite me.

What would you like to achieve in your scientific field?

We conduct fundamental research to contribute to intelligent systems in the long term. This may mean that we advance very small segments or gain elementary insights in underexplored subfields. Specifically, our current goal is to connect previously incompatible fields such as reinforcement learning and causality. We work on all phases of experimental design and we want to automate data collection and make data collection processes significantly more efficient.

I greatly enjoy working in a network of inspiring and dedicated people, and I am very happy and grateful for the opportunities that Helmholtz and TU Munich provide for my research.
Prof. Stefan Bauer

What are the biggest challenges and why is it still worth it every day?

A scientific career is associated with many uncertainties and, above all, instability. Frequent changes of residence are common, and most often, researchers only receive short-term employment contracts. We train many excellent scientists without them ever having the chance of a permanent position in academia. In addition, increasing bureaucratic requirements represent one of the greatest burdens in an already busy work routine. Nevertheless, I greatly enjoy working in a network of inspiring and dedicated people, and I am very happy and grateful for the opportunities that Helmholtz and TU Munich provide for my research.

Was there a formative experience in your scientific career that left mark on you?

I have been greatly influenced by my mentors and their leading example or support. I am very grateful for their support and/or collaboration. And certainly there was a portion of luck.

In your view, what characterizes the life of a scientist?

The continuous endeavor and determination to make an epistemic contribution despite many setbacks and adversities, the desire to have a positive impact with one’s research, sharing of knowledge, and at the same time, the openness to learn at any time and from anyone.
Prof. Stefan Bauer

What do you draw strength from next to your work? What hobbies do you have?

Sports, music, and good food.

Tell a secret about yourself!

I come from the vicinity of Munich, and I am very excited about having returned to my home region.

Prof. Stefan Bauer is an Associate Professor at TU Munich and a Director at Helmholtz AI. He is a CIFAR Azrieli Global Scholar and prior to his appointment in Munich, he was an Assistant Professor at KTH Stockholm and a Group Leader at the Max Planck Institute for Intelligent Systems in Tübingen. Stefan Bauer obtained his Ph.D. in Computer Science from ETH Zurich in 2018, for which he was honored with the ETH Medal for outstanding dissertations. He previously studied Mathematics at ETH Zurich and Economics and Finance at the University of London. His publications have received multiple awards, including the Best Paper Award at the 2019 International Conference on Machine Learning (ICML). As an organizer of competitions for the application and development of AI, such as drug research or real-robot-challenge.com, he advocates for the democratization and transparency of machine learning research.

Latest update: November 2024