Scholar
Tomasz Kacprzak
Google Scholar ID: 1eZr_L0AAAAJ
Swiss Data Science Center, Paul Scherrer Institute, ETH Zurich
Citations & Impact
Citations
8,689
H-index
41
i10-index
73
Publications
20
Co-authors
0
Contact
Publications
Resume (English only)
Academic Achievements
- Paper “Scalable Approximate Algorithms for Optimal Transport Linear Models” under review at Journal of Machine Learning Research (arXiv:2504.04609).
- Paper “SHAM-OT: Rapid Subhalo Abundance Matching with Optimal Transport” published in Monthly Notices of the Royal Astronomical Society Letters (arXiv:2502.17553).
- Paper “DeepLSS: breaking parameter degeneracies in large scale structure with deep learning analysis of combined probes” published in Physical Review X (Phys. Rev. X 12, 031029, 2022), featured in APS Physics Magazine.
- Paper “Laue Indexing with Optimal Transport” under review at IEEE PAMI; software package LaueOT forthcoming on GitHub.
- First deep learning-based cosmological analysis on KiDS-450 dataset published in Phys. Rev. D (2019, 100, 063514) with Janis Fluri, covered by ETH News and MIT Technology Review.
- Introduced optimal transport to multiple applied physics domains and proposed a novel OT-based regression model.
Research Experience
- Former Senior Data Scientist at the Swiss Data Science Center (SDSC), Paul Scherrer Institute.
- Former Senior Scientist at ETH Zurich.
- Leading the collaborative project “Robust and scalable Machine Learning algorithms for Laue 3-Dimensional Neutron Diffraction Tomography” at PSI, developing a novel optimal transport-based indexing method for polycrystalline diffraction patterns.
- Lead Data Scientist for the SDSC–PSI collaboration “Smart Analysis of MUonic x-Rays with Artificial Intelligence”, developing scalable Sinkhorn-like algorithms for muonic X-ray spectral analysis.
- Proposed optimal transport algorithms (SHAM-OT) for matching galaxies and dark matter halos in cosmological simulations.
- Pioneered deep learning approaches for cosmological parameter inference and large-scale structure simulation, achieving 40% improvement in measurement precision.
Co-authors