From AI to climate: Mathematical Institute for Machine Learning and Data Science

The MIDS is a central component in establishing the focus on digitalization at the KU. At the opening in spring 2022, Vice President Prof. Dr. Jens Hogreve emphasized: "We see digitalization as a cross-cutting issue in order to make a contribution to a human-centered digital society. To this end, it is crucial to establish our own expertise in the field of mathematics, which in turn forms the basis for the application of and reflection on data science and artificial intelligence." The MIDS represents added value not only for the University, but also for the region as a location for excellent research and teaching. Accordingly, the city of Ingolstadt has supported the institute with two endowed chairs right from the outset: the Chair of Geomatics and Geomathematics, held by Prof. Dr. Nadja Ray, and the Chair of Applied Mathematics, held by Prof. Dr. Marcel Oliver.

Both are representative of a research focus of the MIDS, namely weather forecasts, climate and soil research. Specifically, it is about processing data to predict such environmental developments. The further development of the mathematical foundations of these methods is highly relevant for a wide range of applications in science and the industry. Another research focus of the institute is the fundamentals of machine learning. Machine learning algorithms are used in areas such as speech and image recognition or self-driving vehicles. "Despite their success in practice, we still lack a comprehensive theoretical understanding of why these methods work so well", explains Prof. Dr. Götz Pfander, spokesperson for the MIDS and holder of the Chair of Mathematics – Computational Science. It has also been repeatedly shown that trained neural networks are often not robust and that even minimal changes to the input can produce an incorrect output. It is essential to analyze the reasons for this instability. "It is of great interest – especially for critical areas of application such as in medicine – to develop improved guarantees of success on a mathematical basis", says Pfander.

The MIDS is also concerned with sustainability, as complex algorithms require immense computing power, which leads to high energy consumption. Tackling this problem is a central goal of the “Resource Aware Artificial Intelligence for Future Technologies” joint project, in which Prof. Dr. Felix Voigtlaender, holder of Chair of Reliable Machine Learning, is involved in collaboration with FAU Erlangen-Nuremberg, TU Munich and the University of Bayreuth. His Chair is funded by the Bavarian Hightech Agenda.

Eight professors and 14 early-career researchers are currently conducting research at MIDS. They also form the core of the Bachelor's degree course in Data Science, which started in the 2022/23 winter semester. The degree program teaches the fundamentals of machine learning and other current methods for data analysis, as well as the ability to efficiently apply these methods using modern software technologies. The English-language Bachelor's degree is also characterized by the international composition of the student body.