Automated Machine Learning for Science
Dr. Katharina Eggensperger is head of the Early Career Research Group ”AutoML for Science” at our Cluster of Excellence ”Machine Learning”.
Research Interests
Machine learning (ML) has become essential to scientific research and modern data-driven applications. However, the application of ML relies on crucial design decisions and technical requirements that demand considerable expertise and resources. We research methods to make the application of ML efficient and easily accessible, aiming at making automated machine learning (AutoML) an inherent part of any (research) project that uses ML methods.
AutoML methods have been demonstrated to be effective for general supervised learning tasks. ML has been successfully used for scientific research, however, specialized AutoML methods are not yet widely available. Through AutoML, we aim to increase the impact and make ML available for new scientific questions. To do so, we want to understand and identify challenges unique to the application of ML in science and leverage existing domain expertise to research new AutoML methods.
The primary focus of this group is to research efficient AutoML systems that, by utilizing machine learning, automatically design predictive pipelines, yield predictions, and insights.
Futher Information
For further information, please visit Katharina Eggensperger's Lab Website.
About Katharina Eggensperger
Katharina Eggensperger joined the Cluster of Excellence "Machine Learning" in Tübingen as an early career research group leader in January 2023. Before, she completed her Ph.D. at the the ML Lab at the University of Freiburg under the supervision of Frank Hutter and Marius Lindauer (2022). Her research focused on methods for AutoML, hyperparameter optimization (HPO), empirical performance modelling, and efficient benchmarking of HPO algorithms. During her time as a doctoral candidate, she co-developed open-source software for HPO methods and AutoML systems and has been a member of the team winning three AutoML competitions (2016, 2018, 2020). Katharina Eggensperger’s research is motivated by the goal of democratizing the application of machine learning for scientific researchers and practitioners.
Contact
Dr. Katharina Eggensperger
AutoML for Science
University Tübingen
Cluster of Excellence "Machine Learning"
Maria-von-Lindenstr. 6, 4th floor
Room No. 40-31/A8
72076 Tübingen
+49 7071 29-70913
katharina.eggensperger@uni-tuebingen.de