Portrait of Dominik Sturm

Dominik Sturm

Position
Wrap-up postdoc, since 2021
Country of origin
Germany
Expertise
Statistical learning, Stochastic generative models, Spatial organization of living systems
Research interest
Statistical learning and stochastic generative models to understand and quantify the spatial organization of living systems.
See CV

Dominik Sturm has been a PhD student in the MOSAIC group between March 2021 and May 2026. Since then, he is a wrap-up postdoc in the group.

Dominik studied Computer Science at the Technische Universität Dresden and obtained his MSc in Computer Science in 2021. In his Master’s thesis titled “Data-Driven Mechanistic Modeling of Biological Processes” at the MOSAIC Group with Dr. Suryanarayana Maddu, Prof. Dr. Ivo F. Sbalzarini and Prof. Dr. Bjoern Andres he investigated strategies to alleviate numerical problems during the training of physics-informed neural networks (PINNs). This enabled the data-driven modeling of turbulence in active fluids by PINNs. The results have been published as part of a journal article in IOP Machine Learning: Science and Technology.

Dominik has several years of professional experience as a Software Engineer. He also was a research assistant with the Fraunhofer Institute for Machine Tools and Forming Technology (IWU), where he developed CNN-based methods for outlier detection in production processes. He also previously worked with the MOSAIC group at CSBD, where he helped with the implementation of several machine-learning projects around data-driven modeling and simulation of living systems.

In our group, Dominik develops statistical learning and stochastic generative models to understand and quantify the spatial organization of living systems.

Publications

  1. Robust variable selection for spatial point processes observed with noise

    D. Sturm and I. F. Sbalzarini

    Spatial Statistics 74:101005, 2026

  2. Spatially Informed Autoencoders for Interpretable Visual Representation Learning

    D. Sturm, H. Bensalem and I. F. Sbalzarini

    International Conference on Learning Representations (ICLR), 2026

  3. Learning Robust and Interpretable Representations of Spatial Point Processes

    D. Sturm

    MOSAIC Group, Faculty of Computer Science, TU Dresden, 2026

  4. Learning locally dominant force balances in active particle systems

    D. Sturm, S. Maddu and I. F. Sbalzarini

    Proc. R. Soc. A 480(2304):20230532, 2024

  5. STENCIL-NET for equation-free forecasting from data

    S. Maddu, D. Sturm, B. L. Cheeseman, C. L. Müller and I. F. Sbalzarini

    Sci. Rep 13:12787, 2023

  6. Inverse Dirichlet weighting enables reliable training of physics informed neural networks

    S. Maddu, D. Sturm, C. L. Müller and I. F. Sbalzarini

    Mach. Learn.: Sci. Technol 3:015026, 2022

  7. Learning computable models from data

    S. Maddu, D. Sturm, B. L. Cheeseman, C. L. Müller and I. F. Sbalzarini

    Proc. 14th World Congress on Computational Mechanics (WCCM), 1–6, 2021

  8. Data-Driven Mechanistic Modeling of Biological Processes

    D. Sturm

    Technische Universität Dresden, Faculty of Computer Science, 2021 (Master thesis)

  9. Dynamic mode decomposition

    D. Sturm

    MOSAIC technical report, MOSAIC Group, TU Dresden, 2019