Learning from Crowds Using a Focal Loss Function: Dealing with Imbalanced Annotations
Technologies · 17.06.2026 · 14 (6) · 370
DOI: 10.3390/technologies14060370 ↗Kurz gefasst
Fachartikel in der Zeitschrift Technologies, online erschienen am 17.06.2026. Verfasst von Julian Gil-Gonzalez und 4 weiteren Autor:innen. Die Arbeit nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministerio de Ciencia, Tecnología e Innovación.
Abstract (Originalsprache)
Obtaining high-quality labeled data for supervised learning is costly, motivating the use of crowdsourcing, which distributes the annotation process across multiple workers with varying levels of expertise. A key challenge in crowdsourced data is annotation sparsity, as each worker labels only a limited subset of instances. This sparsity can amplify class imbalance, reduce supervision for minority classes, and bias standard cross-entropy-based models toward the majority classes. To address this problem, we propose a correlated chained Gaussian process framework trained on a focal-loss-based variational objective (CCGPFL). This probabilistic framework jointly models latent ground-truth and instance-dependent annotator reliability while accounting for correlations among annotators. In addition, the focal-weighted objective mitigates the imbalance induced by sparse annotations by assigning greater importance to harder examples during training. Experiments on synthetic, semi-synthetic, and fully real multi-annotator datasets show that CCGPFL achieves competitive and often superior performance relative to state-of-the-art learning-from-crowds baselines in terms of Overall Accuracy (OA) and Area Under the ROC Curve (AUC).
Förderung
- Ministerio de Ciencia, Tecnología e Innovación
Zitieren
APA
Gil-Gonzalez, J., Cárdenas-Peña, D. A., Orozco-Gutiérrez, A., Guijarro-Estelles, E. D., & Álvarez-Meza, A. M. (2026). Learning from Crowds Using a Focal Loss Function: Dealing with Imbalanced Annotations. Technologies, 14(6), 370. https://doi.org/10.3390/technologies14060370
Vancouver
Gil-Gonzalez J, Cárdenas-Peña DA, Orozco-Gutiérrez A, Guijarro-Estelles ED, Álvarez-Meza AM. Learning from Crowds Using a Focal Loss Function: Dealing with Imbalanced Annotations. Technologies. 2026;14(6):370. doi:10.3390/technologies14060370
BibTeX
@article{GilGonzalez2026,
author = {Gil-Gonzalez, Julian and Cárdenas-Peña, David Augusto and Orozco-Gutiérrez, Alvaro and Guijarro-Estelles, Enrique D. and Álvarez-Meza, Andres M.},
title = {Learning from Crowds Using a Focal Loss Function: Dealing with Imbalanced Annotations},
journal = {Technologies},
year = {2026},
volume = {14},
number = {6},
pages = {370},
doi = {10.3390/technologies14060370}
}
EUVIMED