EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

The SHAP Explainer Model for Binary Classifiers Detecting Functional Groups in Molecules Based on FTIR Spectra

Tomasz Urbańczyk iD, Jakub Bożek iD, Jarosław Koperski iD, Marek Krośnicki iD

International Journal of Molecular Sciences · 20.02.2026 · 27 (4) · 2004

DOI: 10.3390/ijms27042004 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Molecular Sciences, online erschienen am 20.02.2026. Verfasst von Tomasz Urbańczyk und 3 weiteren Autor:innen. Beteiligte Einrichtung: Smoluchowski Institute of Physics, Faculty of Physics, Astronomy and Applied Computer Science, Jagiellonian University, Łojasiewicza 11, 30-348 Krako…. Hauptthemen laut MEDLINE: Deep Learning. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Narodowa Agencja Wymiany Akademickiej, Polish National Agency for Academic Exchange.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

One of the disadvantages of deep learning models is the difficulty in understanding the premises on the basis of which the models make decisions. This might hinder the applicability of the model due to legal issues. In this paper, we investigate the decision process of CNN-KAN model trained to recognize chemical groups based on recorded FTIR spectra. The CNN-KAN model was trained as a binary classifier and using SHAP values, we could trace back the decision-making process and point out which part of the FTIR spectra are responsible for each positive or negative decision. It appears that the decision-making process of the deep learned CNN-KAN model is based on spectral regions that according to the literature have a large impact on the detection of particular functional groups.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningSpectroscopy, Fourier Transform InfraredConvolutional Neural NetworksClassification AlgorithmsAlgorithms

Stichwörter der Autor:innen

FTIR spectraKANexplainable AIfunctional groupsneural networks

Autor:innen und Institutionen

  • Smoluchowski Institute of Physics, Faculty of Physics, Astronomy and Applied Computer Science, Jagiellonian University, Łojasiewicza 11, 30-348 Krakow, Poland.
  • Institute of Theoretical Physics and Astrophysics, Faculty of Mathematics, Physics and Informatics, University of Gdansk, Wita Stwosza 57, 80-308 Gdansk, Poland.

ORCID-Profile: Tomasz Urbańczyk, Jakub Bożek, Jarosław Koperski, Marek Krośnicki

Förderung

  • Narodowa Agencja Wymiany Akademickiej
  • Polish National Agency for Academic Exchange

Zitieren

APA

Urbańczyk, T., Bożek, J., Koperski, J., & Krośnicki, M. (2026). The SHAP Explainer Model for Binary Classifiers Detecting Functional Groups in Molecules Based on FTIR Spectra. International Journal of Molecular Sciences, 27(4), 2004. https://doi.org/10.3390/ijms27042004

Vancouver

Urbańczyk T, Bożek J, Koperski J, Krośnicki M. The SHAP Explainer Model for Binary Classifiers Detecting Functional Groups in Molecules Based on FTIR Spectra. International Journal of Molecular Sciences. 2026;27(4):2004. doi:10.3390/ijms27042004

BibTeX
@article{Urbaczyk2026,
  author = {Urbańczyk, Tomasz and Bożek, Jakub and Koperski, Jarosław and Krośnicki, Marek},
  title = {The SHAP Explainer Model for Binary Classifiers Detecting Functional Groups in Molecules Based on FTIR Spectra},
  journal = {International Journal of Molecular Sciences},
  year = {2026},
  volume = {27},
  number = {4},
  pages = {2004},
  doi = {10.3390/ijms27042004}
}
Kostenlos starten →