Tobias Sutter

I am a (tenure track) assistant professor in machine learning and optimization in the Department of Computer and Information Science at the University of Konstanz. I obtained my PhD degree from the Automatic Control Laboratory at ETH Zurich under the supervision of John Lygeros and was postdoctoral researcher and lecturer at EPFL working with Daniel Kuhn in the Chair of Risk Analytics and Optimization.

E-mail: tobias.sutter@uni-konstanz.de

Curriculum vitae [pdf]

Open positions:

Please contact me via email if you are interested in working with me

News

New paper [arXiv] Towards Optimal Offline Reinforcement Learning
Paper [arXiv] on Computing Optimal Joint Chance Constrained Control Policies accepted to IEEE Transactions on Automatic Control
Two papers at NeurIPS 2024
- Randomized algorithms and PAC bounds for inverse reinforcement learning in continuous spaces [arXiv]
- Newton Losses: Using Curvature Information for Learning with Differentiable Algorithms [arXiv]
New preprint [arXiv] on Randomized algorithms and PAC bounds for inverse reinforcement learning in continuous spaces
New preprint [arXiv] on Probabilistic Verification of Neural Networks using Branch and Bound
New paper [arXiv] on Regularized Q-learning through Robust Averaging (accepted at ICML 2024)
New paper [arXiv] Joint Chance Constrained Optimal Control via Linear Programming, accepted to IEEE Control Systems Letters
New seminar series on Connecting Statistical Logic, Dynamical Systems and Optimization
New preprint [arXiv] on Computing Optimal Joint Chance Constrained Control Policies
Organization of workshop: Moments and Polynomials: Applications and Theory

Teaching (see also Lectures)

Konstanz: Optimization for Data Science, Lecture homepage
Konstanz: Introduction to Machine Learning, Lecture homepage

Publications and Preprints

See the arXiv and Google Scholar or see Publications
PhD thesis [pdf]

PhD Supervision at University of Konstanz

Peter Schmitt-Förster, since 2022
Radek Salač, since 2023
Marcel Kaiser, since 2024
Axel Wolter, since 2024

Research Interests
I am mainly interested in the development of mathematical methodologies and algorithms for optimal data driven decision making. For that I use tools from

Machine learning: reinforcement learning, data-driven decision making,
Control theory: stochastic control, approximate dynamic programming
Stochastic programming: distributionally robust optimization
Statistics: efficient decision making, causal inference,
Information theory: channel capacity approximation, quantum information theory

Awards

ETH medal for outstanding doctoral thesis - 2018 (picture)
IEEE George S. Axelby Outstanding Paper Award - 2016 (picture)