EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Adaptive temperature scaling for Robust calibration of deep neural networks

Sergio A. Balanya, Juan Maroñas, Daniel Ramos

Neural Computing and Applications · 2024-05 · 36 (14) · 8073-8095

DOI: 10.1007/s00521-024-09505-4 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Neural Computing and Applications, online erschienen am 28.02.2024. Verfasst von Sergio A. Balanya und 2 weiteren Autor:innen. Die Arbeit wurde bisher 87-mal zitiert (Crossref) und nennt 44 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministerio de Ciencia e Innovación, Universidad Autónoma de Madrid.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In this paper, we study the post-hoc calibration of modern neural networks, a problem that has drawn a lot of attention in recent years. Despite the plethora of calibration methods proposed, there is no consensus yet on the inherent complexity of the task and, while some authors claim that simple functions solve the problem, others suggest that more expressive models are needed to capture misscalibration. As a first approach, we focus on the task of confidence scaling, specifically on post-hoc methods that generalize Temperature Scaling, which we refer to as the Adaptive Temperature Scaling family. We begin by demonstrating that while complex models like neural networks provide an advantage when there is ample data, they fail in scenarios where it is limited, notably common in fields like medical diagnosis. We then show how under this ideal data conditions the more expressive methods learn a relationship between the entropy of a prediction and its level of overconfidence, and based on this observation, we propose Entropy-based Temperature Scaling, a simple method that scales the confidence of a prediction according to this relationship. Results show that our method obtains state-of-the-art performance and is robust against data scarcity. Moreover, our proposed model enables a deeper understanding of the calibration process by the interpretation of the entropy as a measure of uncertainty in the network outputs.

Förderung

  • Ministerio de Ciencia e Innovación
  • Universidad Autónoma de Madrid

Zitieren

APA

Balanya, S. A., Maroñas, J., & Ramos, D. (2024). Adaptive temperature scaling for Robust calibration of deep neural networks. Neural Computing and Applications, 36(14), 8073-8095. https://doi.org/10.1007/s00521-024-09505-4

Vancouver

Balanya SA, Maroñas J, Ramos D. Adaptive temperature scaling for Robust calibration of deep neural networks. Neural Computing and Applications. 2024;36(14):8073-8095. doi:10.1007/s00521-024-09505-4

BibTeX
@article{Balanya2024,
  author = {Balanya, Sergio A. and Maroñas, Juan and Ramos, Daniel},
  title = {Adaptive temperature scaling for Robust calibration of deep neural networks},
  journal = {Neural Computing and Applications},
  year = {2024},
  volume = {36},
  number = {14},
  pages = {8073--8095},
  doi = {10.1007/s00521-024-09505-4}
}
Kostenlos starten →