Bio
Steffen Schotthoefer is the current Householder Fellow in the Mathematics in Computation Section at Oak Ridge National Laboratory, affiliated with the Multiscale Methods and Dynamics Group.
His research focuses on efficient numerical methods for training and fine-tuning AI models at scale and under resource constraints. This includes low-rank methods for model compression that reduce the computational cost of neural network training and inference, as well as neural network-based surrogate models for scientific applications such as radiation transport and plasma dynamics.
Steffen earned his PhD in Applied Mathematics from KIT, Germany, where he developed surrogate modeling techniques for radiation transport and numerical methods for kinetic PDEs.
Education
- Karlsruhe Institute of Technology, Germany, Ph.D. Mathematics, 2023
- TU Kaiserslautern, Germany, M.Sc. Mathematics and Computer Science, 2020
- TU Kaiserslautern, Germany, B.Sc. Mathematics, 2017
Professional Service
- Reviewer for NeurIPS, ICML, SIAM
- Organizer of the International Workshop on Moment Methods in Kinetic Theory IV
Professional Affiliations
Publications
Structure-preserving neural networks for the regularized entropy-based closure of a linear, kinetic…
Other Publications
See Google Scholar