Department of Mathematics and Computer Science · Saarland University
Mathematical Optimization for Data Science Group
We develop and analyze optimization algorithms for data science and machine learning: non-smooth and non-convex optimization, inertial and accelerated methods, bilevel optimization and automatic differentiation, dynamical systems, and learnable optimization algorithms with provable guarantees.
Group leader: Prof. Dr. Peter Ochs
News
All news →- Oct 2026We are looking forward to the 2026-edition of our yearly MOP Retreat at Saarland University starting on October 26th.
- Oct 2026We have two new preprints Euclidean Distance-Based Flow Matching and Optimal Transport PAC-Bayes Bounds on arXiv.
- Sep 2026Our paper Understanding the Curse of Unrolling was accepted at NeurIPS.
- Aug 2026We have a new preprint A Probabilistic Framework for Learnable Optimization Algorithms on arXiv.
- Apr 2026Our paper Symmetries in PAC-Bayesian Learning was accepted at ICML.
- Apr 2026Congratulations to Shida on his graduation!
- Apr 2026Mahesh Chandra Mukkamala joins our group as a PostDoc.
- Nov 2025Paul Heilmann joins our group as a PhD Student.
- Oct 2025We have a new preprint Symmetries in PAC-Bayesian Learning on arXiv.
- Oct 2025We have a new preprint Bregman Stochastic Proximal Point Algorithm with Variance Reduction on arXiv.
