EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

DEEP NON-SMOOTH REGRESSION FOR BIOLOGICAL STRUCTURES

DENYS CHERHYKALO

Modeling Control and Information Technologies · 11.11.2026 · (9) · 85

DOI: 10.31713/mcit.2026.012 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Modeling Control and Information Technologies, online erschienen am 11.11.2026. Verfasst von DENYS CHERHYKALO. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

In many cases, existing attractors can be described by a bifurcation cascade. This information can be used to construct a neural network in which each successive layer models the next stage of the bifurcation cascade, until the desired structure and the associated objective function values ​​are obtained. Materials and Methods. If the regression criteria are not standard but involve first- and second-order derivatives, attempts to iteratively approach the desired solution may result in significant noise—particularly if the solution exhibits non-smoothness. The solution lies in understanding the nature of this solution and the causes of its non-smoothness. If the function in question reflects system behavior characterized by a specific balance dependent on initial conditions, the non-smoothness most likely arises from the presence of multiple attractors—and, consequently, multiple system regimes.

Results

This paper develops a physics-informed neural network model that describes the cascade of bifurcations in biological systems using the theory of smooth approximation.

Conclusions

This approach makes it possible to seek approximations for the parameters of these hidden bifurcation cascades themselves—parameters that can initially be smoothed and then gradually restored to a non-smooth form during the course of iterations. The paper cites, as an example, functions representing the properties of various cell types, which are essentially stable attractors formed through bifurcations of cell fates.

Zitieren

APA

CHERHYKALO, D. (2026). DEEP NON-SMOOTH REGRESSION FOR BIOLOGICAL STRUCTURES. Modeling Control and Information Technologies, 85. https://doi.org/10.31713/mcit.2026.012

Vancouver

CHERHYKALO D. DEEP NON-SMOOTH REGRESSION FOR BIOLOGICAL STRUCTURES. Modeling Control and Information Technologies. 2026:85. doi:10.31713/mcit.2026.012

BibTeX
@article{CHERHYKALO2026,
  author = {CHERHYKALO, DENYS},
  title = {DEEP NON-SMOOTH REGRESSION FOR BIOLOGICAL STRUCTURES},
  journal = {Modeling Control and Information Technologies},
  year = {2026},
  number = {9},
  pages = {85},
  doi = {10.31713/mcit.2026.012}
}
Kostenlos starten →