Prof. Dr. Holger Hoos
Alexander von Humboldt-Professor of Computer Science of the Faculty of Mathematics, Informatics and Natural Sciences at RWTH Aachen University
»I see great potential in the strong networking of AI expertise and AI activities, not only in academic research, but also in the training of AI talent and in applications in the industry.«
Prof. Dr. Holger Hoos is Alexander von Humboldt Professor since 01.01.2022 for Computer Science of the Faculty of Mathematics, Informatics and Natural Sciences at RWTH Aachen University. The Alexander von Humboldt Professorship is Germany’s most highly endowed international research award. The official award ceremony took place in Berlin on 12th May. In a short interview, we asked Prof. Hoos about his field of expertise, so-called Automated Machine Learning (AutoML). We also wanted to know what he wants to achieve in his new position as Humboldt Professor.
How would you explain Automated Machine Learning (AutoML) to a non-specialist?
The goal of AutoML is to select and calibrate machine learning (ML) algorithms and systems automatically for specific use cases. In doing so, we enable people with limited ML expertise to use such algorithms and systems effectively and responsible, and for ML experts to achieve better results with less effort.
What is the biggest achievement in AutoML so far?
Just as for AI as a whole, or the field of ML, this question cannot be answered unambiguously for AutoML. AutoML methods are increasingly used in research and also in business; nevertheless, this is still a rather new approach, whose greatest successes will only become apparent in the next few years. However, already in a competition in 2016, it was convincingly demonstrated that AutoML systems can achieve better results than human experts, e. g., in the use of ML methods in the development of new drugs and active ingredients.
What are your (future) goals for AutoML? What will we be facing in the future in the AutoML area?
My work is not only about basic research and applications in the area of AutoML, but also AI is much more than just ML, just as human intelligence is not only based on the ability to learn, but also, for example, on the ability to plan and reason. The extension of AutoML to AI as a whole is called »Automated Artificial Intelligence« or AutoAI for short. In my view, it represents an essential component for the development of the next generation of AI systems. Increasingly, this is not only about the performance of AI systems, but also about other important properties, such as robustness to interference or attempts at manipulation, as well as energy consumption.
How can AutoML be used specifically in companies in NRW?
For companies that employ ML experts, primarily by having these experts integrate AutoML methods and -tools into their workflows. Companies that do not have in-house ML expertise should be supported by experts with AutoML knowledge; this can also be done, for example, through cooperations with university research institutes. In principle, the use of AutoML and AutoAI methods is of interest to all companies that already use or want to use ML and AI methods.
As Distinguished Scientific Director of the AI Center at RWTH Aachen University, you are responsible for the foundation of a trinational research network between Belgium, the Netherlands and Germany. What form could have a cooperation between these three research locations?
I see great potential in the strong networking of AI expertise and AI activities, not only in academic research, but also in the training of AI talent and in applications in the industry. Specifically, the top universities in the three countries should cooperate more in this area, both in research and in teaching. Similarly, there should be greater trinational networking among companies and other users (e. g., in the public sector), for example in the form of user forums, continuing education offerings and talent fairs. There are already good local and regional and pan-European networks such as CLAIRE (claire-ai.org) offer further possibilities to get something like this off the ground quickly and effectively. Especially Aachen, which is located geographically and historically in the heart of Europe, is an excellent nexus for this project to establish stronger connections between three European core states with a high level of AI expertise. We will all benefit from this.
You have been awarded the title of Humboldt Professor. What do you want to achieve in your new position?
I have three main goals:
- to advance the next generation of AI methods and systems in basic research, with a focus on AI that complements rather than seeks to replace human intelligence.
- through research and applications in the field of AutoAI, to promote a broad, effective and responsible use of AI methods, particularly in key areas such as medicine & health or sustainability & climate change.
- contribute to the global success of »AI made in Europe« by helping to strengthen Germany as a technology location and by engaging at the European level.
Since 01. January 2022 Prof. Dr. Holger Hoos is Alexander von Humboldt Professor of Artificial Intelligence at the Faculty of Mathematics, Computer Science and Natural Sciences at RWTH Aachen University. In addition, he continues to work part-time at the Leiden Institute of Advanced Computer Science (LIACS) at Leiden University in the Netherlands, where he has headed the Chair of Machine Learning since 2016. He is a fellow of the Association for Computing Machinery (ACM), the Association for the Advancement of Artificial Intelligence (AAAI), and the European Association for Artificial Intelligence (EurAI). Prof. Hoos is co-founder and chairman of the board of directors of the Confederation of Laboratories for AI Research in Europe (CLAIRE), which is committed to strengthening AI research in Europe and was awarded the German AI Innovation Prize in 2021.