معرفی
Timo Kröger is a Researcher at the University of Münster's Institute for Analysis and Numerics, part of the Department of Mathematics and Computer Science. His research focuses on numerical analysis, partial differential equations, and the mathematical foundations of artificial neural networks. He collaborates with Prof. Arnulf Jentzen on projects exploring the dynamics of neural network training, optimization landscapes, and high-dimensional approximation problems.
Research interests include the theoretical analysis of gradient descent methods in overparameterized networks, non-convex optimization challenges in deep learning, and Lipschitz continuity properties of neural network parameters. His work often intersects with applied analysis and computational mathematics.
Publications: His recent studies investigate the existence of non-global minima in neural network training (2022), parameter norm bounds via Lipschitz analysis (2022), and convergence rates for gradient descent in overparameterized systems (2021).
Professional Support: Administrative support is provided by Claudia Giesbert (Phone: +49 251 83-33792, Room 120.002). Kröger's office is located in Room 110.025 at Orleans-Ring 10, 48149 Münster.
