EUVIMEDEUVIMED entdecken
Open Access

Rethinking Cost-Sensitive Classification in Deep Learning via Adversarial Data Augmentation

Qiyuan Chen iD, Raed Al Kontar iD, Maher Nouiehed iD, X Jessie Yang iD, Corey Lester iD

INFORMS Journal on Data Science · 2025-01 · 4 (1) · 1-19

DOI: 10.1287/ijds.2022.0033 ↗

Kurz gefasst

Fachartikel in der Zeitschrift INFORMS Journal on Data Science, 2025-01. Verfasst von Qiyuan Chen und 4 weiteren Autor:innen. Beteiligte Einrichtung: Industrial & Operation Engineering, University of Michigan, Ann Arbor, Michigan 48109. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NLM NIH HHS.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Cost-sensitive classification is critical in applications where misclassification errors widely vary in cost. However, overparameterization poses fundamental challenges to the cost-sensitive modeling of deep neural networks (DNNs). The ability of a DNN to fully interpolate a training data set can render a DNN, evaluated purely on the training set, ineffective in distinguishing a cost-sensitive solution from its overall accuracy maximization counterpart. This necessitates rethinking cost-sensitive classification in DNNs. To address this challenge, this paper proposes a cost-sensitive adversarial data augmentation (CSADA) framework to make overparameterized models cost sensitive. The overarching idea is to generate targeted adversarial examples that push the decision boundary in cost-aware directions. These targeted adversarial samples are generated by maximizing the probability of critical misclassifications and used to train a model with more conservative decisions on costly pairs. Experiments on well-known data sets and a pharmacy medication image (PMI) data set, made publicly available, show that our method can effectively minimize the overall cost and reduce critical errors while achieving comparable overall accuracy. History: Nick Street served as the senior editor for this article. Funding: Research reported in this publication was supported by the National Library of medicine of the National Institutes of Health in the United States under award number R01LM013624. Data Ethics & Reproducibility Note: This paper abides by data ethics requirements. Data used are publicly available online at https://deepblue.lib.umich.edu/data/concern/data_sets/6d56zw997 . Codes to replicate the results of this paper are available on Code Ocean at https://doi.org/10.24433/CO.2139841.v1 and in the e-Companion to this article (available at https://doi.org/10.1287/ijds.2022.0033 ).

Schlagwörter

Stichwörter der Autor:innen

adversarial data augmentationcost-sensitive learningdeep neural networksmulticlass classificationoverparametrization

Autor:innen und Institutionen

  • Industrial & Operation Engineering, University of Michigan, Ann Arbor, Michigan 48109.
  • Industrial Engineering, American University of Beirut, Beirut 1107 2020, Lebanon.
  • College of Pharmacy, University of Michigan, Ann Arbor, Michigan 48109.

ORCID-Profile: Qiyuan Chen, Raed Al Kontar, Maher Nouiehed, X Jessie Yang, Corey Lester

Förderung

  • NLM NIH HHS

Zitieren

APA

Chen, Q., Al Kontar, R., Nouiehed, M., Yang, X. J., & Lester, C. (2025). Rethinking Cost-Sensitive Classification in Deep Learning via Adversarial Data Augmentation. INFORMS Journal on Data Science, 4(1), 1-19. https://doi.org/10.1287/ijds.2022.0033

Vancouver

Chen Q, Al Kontar R, Nouiehed M, Yang XJ, Lester C. Rethinking Cost-Sensitive Classification in Deep Learning via Adversarial Data Augmentation. INFORMS Journal on Data Science. 2025;4(1):1-19. doi:10.1287/ijds.2022.0033

BibTeX
@article{Chen2025,
  author = {Chen, Qiyuan and Al Kontar, Raed and Nouiehed, Maher and Yang, X Jessie and Lester, Corey},
  title = {Rethinking Cost-Sensitive Classification in Deep Learning via Adversarial Data Augmentation},
  journal = {INFORMS Journal on Data Science},
  year = {2025},
  volume = {4},
  number = {1},
  pages = {1--19},
  doi = {10.1287/ijds.2022.0033}
}
Kostenlos starten →