Sebastian Kaltenbach

Sebastian Kaltenbach is a researcher at the Technical University of Munich (TUM) working in the Department of Mechanical Engineering under Prof. Faidon-Stelios Koutsourelakis in the Professorship for Data-driven Materials Modeling. He teaches courses on Modeling of Uncertainties and Data in Mechanical Engineering and supervises numerous student projects including Master's theses, Bachelor's theses, and research internships.

His research focuses on the intersection of machine learning and computational physics, particularly in developing physics-aware probabilistic models for multiscale systems. Kaltenbach specializes in incorporating physical constraints into deep learning frameworks, coarse-graining of dynamical systems, uncertainty quantification, and model-order reduction in the small data regime. His work addresses fundamental challenges in creating data-driven models that respect physical laws while requiring minimal training data.

Analyzing his publication record from 2019-2021 reveals a clear trajectory toward increasingly sophisticated physics-informed machine learning approaches. His research consistently addresses the challenge of working with limited data in complex physical systems, with a strong emphasis on uncertainty quantification and stability guarantees. The publications demonstrate expertise across multiple venues including top conferences like ICLR, NeurIPS, and SIAM meetings, covering applications from medical diagnostics to mechanical engineering problems.

Kaltenbach has successfully supervised numerous graduate students on projects spanning neural networks for system identification, Bayesian GANs for medical image synthesis, machine learning of ordinary differential equations, and physics-based neural networks for predicting dynamic systems. His collaborations extend to industry partners like BMW and academic collaborators at institutions including ETH Zurich.