Andrei Caragea

Andrei Caragea is a Researcher at the Mathematical Institute for Machine Learning and Data Science (MIDS) within the Faculty of Mathematics and Geography at Katholische Universität Eichstätt-Ingolstadt. His work bridges Gabor analysis, exponential bases, and neural network approximation theory, with side interests in number theory and classical geometry.

• Education: PhD in Mathematics (2023), Katholische Universität Eichstätt-Ingolstadt
• Key Collaborators: Prof. Götz Pfander, Felix Voigtlaender, Friedrich Philipp, Dae Gwan Lee

His research explores structural limitations in time-frequency representations (Balian-Low theorems for subspaces) and dimension-independent approximation capabilities of complex-valued neural networks. Recent work demonstrates how modReLU-activated networks can overcome the curse of dimensionality for Barron-class classifiers.

Publication Trends (2019-2023):

• 50% focus on Gabor systems and time-frequency invariance
• 40% on neural network expressivity with complex-valued architectures
• 10% on exponential basis constructions in signal processing

Research Grants:

• DFG Project (2019-2025): Sampling theory and bases from exponential functions
• DFG Project (2015-2022): Covariance matrix estimation under sparsity