AdaptiveSSL: a unified semi-supervised learning framework for robust classification via adaptive uncertainty calibration and dynamic labeling
Knowledge and Information Systems · 2026-12 · 68 (1)
DOI: 10.1007/s10115-025-02634-9 ↗Kurz gefasst
Fachartikel in der Zeitschrift Knowledge and Information Systems, online erschienen am 06.01.2026. Verfasst von Usman Gani Joy und 1 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 51 Literaturangaben.
Abstract (Originalsprache)
In industries like telecommunications and banking, scarce labeled data and high misprediction costs challenge traditional machine learning. Semi-supervised learning (SSL) and active learning (AL) offer solutions but often suffer from overconfident predictions, redundant sampling, and static pseudo-labeling thresholds. Existing methods tackle these issues separately, limiting their real-world impact. We introduce AdaptiveSSL, a scalable SSL framework that unifies Wasserstein-based uncertainty calibration, diversity-driven sampling via multi-resolution hashing, and dynamic pseudo-labeling through an adaptive, performance-driven thresholding mechanism. By iteratively refining confidence, ensuring diverse sampling, and dynamically adjusting thresholds based on a composite objective function, AdaptiveSSL achieves robust classification. On the Sparkify churn dataset, it reaches an AUC of 0.9326 with 10% labeled data, outperforming baselines by up to 10% while maintaining efficiency. On the imbalanced Home Credit dataset, it detects the minority class with an TPR score of 0.4521, surpassing model like FreeMatch. Tested on five datasets, AdaptiveSSL provides a practical tool for data-efficient, high-stakes classification.
Zitieren
APA
Joy, U. G., & Rahman, M. M. (2026). AdaptiveSSL: a unified semi-supervised learning framework for robust classification via adaptive uncertainty calibration and dynamic labeling. Knowledge and Information Systems, 68(1). https://doi.org/10.1007/s10115-025-02634-9
Vancouver
Joy UG, Rahman MM. AdaptiveSSL: a unified semi-supervised learning framework for robust classification via adaptive uncertainty calibration and dynamic labeling. Knowledge and Information Systems. 2026;68(1). doi:10.1007/s10115-025-02634-9
BibTeX
@article{Joy2026,
author = {Joy, Usman Gani and Rahman, M. M.},
title = {AdaptiveSSL: a unified semi-supervised learning framework for robust classification via adaptive uncertainty calibration and dynamic labeling},
journal = {Knowledge and Information Systems},
year = {2026},
volume = {68},
number = {1},
doi = {10.1007/s10115-025-02634-9}
}
EUVIMED