Nonlinear Dimensionality Reduction With Diffusion Maps: Guidance for Practical Applications and Interpretation
IEEE Access · 2026 · 14 · 164823-164844
DOI: 10.1109/access.2026.3746320 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Sönke Beier und 4 weiteren Autor:innen. Die Arbeit nennt 79 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The Diffusion Map method is a spectral dimensionality reduction technique which is able to uncover nonlinear manifolds in high-dimensional data. It is increasingly applied across a wide range of scientific disciplines, such as biology, engineering, and social sciences. But data preprocessing, parameter settings and component selection have a significant influence on the resulting manifold, something which has not been comprehensively discussed in the literature so far. Additionally, how to identify the most relevant diffusion components remains an open question, as the leading components are not necessarily the most meaningful and the optimal number of components to retain remains unclear. Here we address these challenges, by numerically investigating the synthetic Swiss roll dataset and the real-world V-Dem democracy dataset. We provide a practice-oriented review of the Diffusion Map technique, highlight common pitfalls and show the effects of parameter settings and data preprocessing. Finally, we introduce a novel, quantitative neural network-based method for component selection.
Zitieren
APA
Beier, S., Pirker-Díaz, P., Pagenkopf, F., Ersoy, I. T., & Wiesner, K. (2026). Nonlinear Dimensionality Reduction With Diffusion Maps: Guidance for Practical Applications and Interpretation. IEEE Access, 14, 164823-164844. https://doi.org/10.1109/access.2026.3746320
Vancouver
Beier S, Pirker-Díaz P, Pagenkopf F, Ersoy IT, Wiesner K. Nonlinear Dimensionality Reduction With Diffusion Maps: Guidance for Practical Applications and Interpretation. IEEE Access. 2026;14:164823-164844. doi:10.1109/access.2026.3746320
BibTeX
@article{Beier2026,
author = {Beier, Sönke and Pirker-Díaz, Paula and Pagenkopf, Friedrich and Ersoy, Ibrahim Talha and Wiesner, Karoline},
title = {Nonlinear Dimensionality Reduction With Diffusion Maps: Guidance for Practical Applications and Interpretation},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {164823--164844},
doi = {10.1109/access.2026.3746320}
}
EUVIMED