Portrait of Paulo von Petersenn

Paulo von Petersenn

Position
PhD student, since 2025
Country of origin
Germany
Expertise
Dynamical systems view of neural networks, Sparse learning, Interpretable latent dynamics
Research interest
A high-dimensional dynamical systems perspective on neural networks: sparse learning, and the inference of interpretable latent dynamics from high-dimensional neural representations.
See CV

Paulo von Petersenn joined the MOSAIC group as a PhD student in October 2025. He comes from Germany.

He earned his B.Sc. in Physics and M.Sc. in Mathematical Physics at the University of Leipzig where he worked on cancer-related biophysics and spent a DAAD-funded year at the National Taiwan University studying quantum information theory.

In his Master’s thesis under the supervision of Prof. Dr. Jürgen Jost at the Max Planck Institute for Mathematics in the Sciences, he explored the relationship between information-theoretic principles and practical L0 regularization techniques, using compression as a unifying lens to bridge theoretical insights with algorithmic implementation.

For his PhD project in the MOSAIC Group, Paulo works on a high-dimensional dynamical systems perspective on neural networks. He investigates methods of sparse learning and for inferring interpretable latent dynamics from high-dimensional neural representations.