We work on the statistical theory of machine learning and deep learning. We provide mathematical understanding of learning algorithms, thereby establishing black-box AI tools as formal statistical principles.

We currently focus on the following topics.

• Statistical theory of unsupervised learning
• Theory of deep learning
• Machine learning on graphs
• Non-parametric methods for learning

We regularly offer lecture/seminars related to the theory of machine learning, including:

• Statistical foundations of learning (lecture, held every summer semester)
• Theoretical advances in deep learning (seminar, typically held every summer semester)
• Recent advances in theoretical machine learning (reading group, held throughtout the year and open for everyone)

We also supervise projects / theses on theoretical and applied machine learning. See here for recent projects.