EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Feature attribution methods in machine learning: a state-of-the-art review

Saidjon Kamolov

Annals of Mathematics and Computer Science · 26.08.2025 · 29 · 104-111

DOI: 10.56947/amcs.v29.635 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Annals of Mathematics and Computer Science, online erschienen am 26.08.2025. Verfasst von Saidjon Kamolov. Die Arbeit wurde bisher 9-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents a state-of-the-art survey of feature attribution techniques employed in explainable AI. We organize the existing literature into a proposed taxonomy of model-agnostic and model-specific approaches. We analyze the formal definitions, mathematical formulations, usage contexts, strengths, and limitations of these methods. A comparative analysis highlights key trade-offs concerning model agnosticism, explanation form, computational cost, and fidelity to the model. We find that while model-agnostic techniques offer broad applicability by treating models as oracles, often at a higher computational cost, model-specific methods leverage internal model architecture or gradients for potentially more efficient and faithful explanations, albeit with reduced generality.

Zitieren

APA

Kamolov, S. (2025). Feature attribution methods in machine learning: a state-of-the-art review. Annals of Mathematics and Computer Science, 29, 104-111. https://doi.org/10.56947/amcs.v29.635

Vancouver

Kamolov S. Feature attribution methods in machine learning: a state-of-the-art review. Annals of Mathematics and Computer Science. 2025;29:104-111. doi:10.56947/amcs.v29.635

BibTeX
@article{Kamolov2025,
  author = {Kamolov, Saidjon},
  title = {Feature attribution methods in machine learning: a state-of-the-art review},
  journal = {Annals of Mathematics and Computer Science},
  year = {2025},
  volume = {29},
  pages = {104--111},
  doi = {10.56947/amcs.v29.635}
}
Kostenlos starten →