The Chair of Bioanalytics and Data Science at the University of Potsdam (joint appointment with Fraunhofer IZI-BB) develops robust methods for pattern recognition and feature extraction from biomedical datasets.
Our research focuses on unsupervised learning, explainable AI, data fusion, as well as mathematical modeling and optimization. Methodological work is directly applied to bioanalytical applications such as point-of-care diagnostics or predictive biomarker identification.
Location 1: University of Potsdam, Am Mühlenberg 9, Haus 62 (H-Lab),
Room 02.62.1.04
14476 Potsdam – Golm
Location 2: Fraunhofer IZI-BB, Am Mühlenberg 13, 14476 Potsdam – Golm
We use matrix and tensor decompositions for clustering, pattern recognition, and feature extraction. Additional constraints such as non-negativity or temporal regularity, improve interpretability and reduce noise. These methods can also enhance explainability in supervised learning.
The group addresses the entire process chain from modeling and optimization to application. For complex and large datasets, we develop tailored models for unsupervised learning and data fusion. This yields high-dimensional optimization problems with many (often nonsmooth) constraints, necessitating the advancement of algorithms on smooth (matrix) manifolds.
Our methods are applied to clustering cancer patients via integrative analysis of multi-omics data and to pattern recognition in time-series data, as well as to analysis and feature extraction for data generated by point-of-care diagnostic tools. Furthermore, the data structures can be used in quantum chemistry calculations and signal processing.
M. Bachmayr, S. Krämer, M. Pfeffer:
Low-rank eigenvalue solvers for block-sparse matrix product states. (2026)
[bibtex],[arxiv]
V. Zalbertus, M. Pfeffer, A. Schmeding:
Optimization on Weak Riemannian Manifolds. (2026)
[bibtex],[arxiv]
V. Borovik, H. Friedman, S. Hoşten, M. Pfeffer:
Numerical Algebraic Geometry for Energy Computations on Tensor Train Varieties. (2025)
[bibtex],[arxiv]
R. Bergmann, H. Jasa, P. John, M. Pfeffer:
The Intrinsic Riemannian Proximal Gradient Method for Convex Optimization. (2025)
[bibtex],[arxiv]
R. Bergmann, H. Jasa, P. John, M. Pfeffer:
The Intrinsic Riemannian Proximal Gradient Method for Nonconvex Optimization. (2025)
[bibtex],[arxiv]
K. Kour, S. Dolgov, P. Benner, M. Stoll, M. Pfeffer:
A weighted subspace exponential kernel for support tensor machines. (2023)
[bibtex],[arxiv]
C. Chatzis, C. Schenker, M. Pfeffer, E. Acar:
tPARAFAC2: Tracking evolving patterns in (incomplete) temporal data. (2025)
[bibtex],[link],[arxiv]
J. Koenig, M. Pfeffer, M. Stoll:
Efficient training of Gaussian processes with tensor product structure.
Computational Optimization and Applications (2025)
[bibtex],[link],[arxiv]
F. Reggiani, Z. El Rashed, M. Petito, M. Pfeffer, A. Morabito, E. T. Tanda, F. Spagnolo, M. Croce, U. Pfeffer, A. Amaro:
Machine Learning Methods for Gene Selection in Uveal Melanoma.
International Journal of Molecular Sciences 25(3) (2024)
[bibtex],[link]
M. Pfeffer, J. Samper:
The cone of 5×5 completely positive matrices.
Discrete & Computational Geometry (2024)
[bibtex],[link],[arxiv]
A. Amaro, M. Pfeffer, U. Pfeffer, F. Reggiani:
Evaluation and Comparison of Multi-Omics Data Integration Methods for Subtyping of Cutaneous Melanoma.
Biomedicines 10(12) (2022)
[bibtex],[link]
H. Eisenmann, F. Krahmer, M. Pfeffer, A. Uschmajew:
Riemannian thresholding methods for row-sparse and low-rank matrix recovery.
Numerical Algorithms (2022)
[bibtex],[link],[arxiv]
M. Bachmayr, M. Götte, M. Pfeffer:
Particle number conservation and block structures in Matrix Product States.
Calcolo 59, 24 (2022)
[bibtex],[link],[arxiv]
C. Krumnow, M. Pfeffer, A. Uschmajew:
Computing eigenspaces with low rank constraints.
SIAM Journal on Scientific Computing, 43 (2021) 1, p. 586-608
[bibtex],[link],[preprint]
M. Eigel, M. Marschall, M. Pfeffer, R. Schneider:
Adaptive Stochastic Galerkin FEM for lognormal coefficients in hierarchical tensor representations.
Numerische Mathematik, 145 (2020) 3, p. 655-692
[bibtex],[link],[arxiv]
M. Pfeffer, A. Seigal, B. Sturmfels:
Learning paths from signature tensors.
SIAM journal on matrix analysis and applications, 40 (2019) 2, p. 394-416
[bibtex],[link],[arxiv],[github]
M. Pfeffer, A. Uschmajew, A. Amaro, U. Pfeffer:
Data fusion techniques for the integration of multi-domain genomic data from uveal melanoma.
Cancers, 11 (2019) 10, 1434
[bibtex],[link],[preprint]
M. Eigel, M. Pfeffer, R. Schneider:
Adaptive stochastic Galerkin FEM with hierarchical tensor representations.
Numerische Mathematik, 136 (2017) 3, p. 765-803
[bibtex],[link],[preprint]
S. Szalay, M. Pfeffer, V. Murg, G. Barcza, F. Verstraete, R. Schneider, Ö. Legeza:
Tensor product methods and entanglement optimization for ab initio quantum chemistry.
International journal of quantum chemistry, 115 (2015) 19, p. 1342-1391
[bibtex],[link],[arxiv]
C. Chatzis, M. Pfeffer, P. Lind, E. Acar:
A Time-aware tensor decomposition for tracking evolving patterns.
2023 IEEE 33rd International Workshop on Machine Learning for Signal Processing (MLSP)
[bibtex],[link],[arxiv]
M. Pfeffer: Tensor methods for the numerical solution of high-dimensional parametric partial differential equations.
Dissertation, Technische Universität Berlin, 2018
[bibtex],[link]
M. Pfeffer: Aspects of second-order optimization on fixed rank tensor manifolds.
Masterarbeit, Technische Universität Berlin, 2015
M. Pfeffer: Dynamical low rank approximation in novel TT format.
Bachelorarbeit, Technische Universität Berlin, 2011
