Johannes Maly

Johannes Maly is an Assistant Professor at the Bavarian AI Chair for Mathematical Foundations of Artificial Intelligence at LMU Munich. He previously held postdoctoral positions at Catholic University of Eichstaett-Ingolstadt and RWTH Aachen University, and completed his PhD at TUM Munich under Prof. Massimo Fornasier.

• PhD in Mathematics (2019, TUM Munich)
• M.Sc. in Mathematics (2015, TUM Munich)
• B.Sc. in Mathematics (2013, TUM Munich)

His research focuses on mathematical data science and machine learning, specifically addressing:

• Robust covariance estimation under quantization
• Neural network approximation properties
• Implicit bias in gradient descent training
• Multi-structured signal recovery
• Quantization effects in deep learning and compressed sensing

His recent publications analyze dithered quantization in covariance estimation, implicit regularization in overparameterized models, and multi-structured data recovery. He applies mathematical rigor to practical challenges in wireless communications (e.g., MIMO systems) and neural network training.

Scientific recognition includes:

• relAI Fellow
• MCML Associate

He supervises code/toolbox development for reproducibility and teaches graduate courses in convex optimization, high-dimensional probability, and mathematical data science. His work bridges theoretical mathematics and applied signal processing.