Prof. Dr. Peter Ochs

Research interests

• Non-smooth analysis and optimization (convex and non-convex)
• Learning of optimization algorithms
• Mathematics of generative models in machine learning
• Algorithms in machine learning, computer vision, inverse problems, and statistics
• Bilevel optimization / parameter learning and automatic differentiation
• Dynamical systems perspective on optimization algorithms

Peter Ochs received the M.Sc. degree in mathematics from Saarland University in 2010 and the Ph.D. degree in mathematics from the University of Freiburg in 2015, including a research stay at TU Graz, Austria. After a postdoctoral position at Saarland University and a return to Freiburg, he was appointed Juniorprofessor of Applied Mathematics at Saarland University in November 2017. In September 2020, he became Tenure-Track Professor at the interface of Mathematics and Computer Science at the University of Tübingen, where he was a full member of the Cluster of Excellence Machine Learning and associate faculty of the IMPRS-IS graduate school. Following a positive tenure evaluation in Tübingen, he accepted a full professorship for Mathematics and Computer Science with a focus on Mathematical Data Analysis at Saarland University in March 2023, where he heads the Mathematical Optimization for Data Science group. He is faculty of the IMPRS-Trust graduate school at the Max Planck Institute for Informatics in Saarbrücken.

His research focuses on non-smooth and non-convex optimization for machine learning, data analysis, computer vision, and inverse problems, with an emphasis on rigorous algorithmic guarantees and the integration of statistical learning principles into optimization methods.

Awards

• Gold Reviewer Award at ICML 2026
• Oral presentation at ICML 2025: A Generalization Result for Convergence in Learning-to-Optimize
• Oral presentation at ECCV 2018: Lifting Layers: Analysis and Applications
• Best Paper Award at GCPR 2016: FSI Schemes: Fast Semi-Iterative Solvers for PDEs and Optimisation Methods
• Best Paper Award at SSVM 2015: Bilevel Optimization with Nonsmooth Lower Level Problems
• Outstanding Reviewer Award at CVPR 2016 and ECCV 2014