CNN-Based Classification of Ziziphus Seeds with Focal Loss for Overcoming Size-Based Shortcut Learning
Biosensors · 02.06.2026 · 16 (6) · 320
DOI: 10.3390/bios16060320 ↗Kurz gefasst
Fachartikel in der Zeitschrift Biosensors, online erschienen am 02.06.2026. Verfasst von Yea-Jin Park und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Smart Farm Science, Kyung Hee University, Yongin 17104, Republic of Korea. Hauptthemen laut MEDLINE: Ziziphus, Seeds. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Herbal medicines represent a significant global market, yet food safety remains threatened by counterfeit products morphologically resembling authentic samples. Models trained on limited datasets are prone to shortcut learning, relying on superficial features rather than intrinsic morphological characteristics. This study identified size-based shortcut learning as a critical factor degrading the classification of Ziziphus jujuba Mill. var. spinosa and its counterfeit Ziziphus mauritiana Lam., and demonstrated that focal loss alone can effectively mitigate this issue. Models trained on the internal dataset were evaluated on an external dataset acquired with the Herb-X. On the internal test set, all configurations achieved high classification accuracies (≥98%), thereby obscuring meaningful differences in external generalization. However, consistent performance degradation was observed on the external dataset. The cross-entropy model trained on background-removed data dropped to 82.08 ± 10.97%, while size-normalized models recovered to 84.17 ± 10.15% (upsizing) and 88.94 ± 6.76% (downsizing), confirming that suppressing size shortcuts improves external generalization. The focal loss model, without any preprocessing, achieved 90.88 ± 2.71%, reducing the internal–external generalization gap from 16.18 to 8.11 percentage points. Grad-CAM++ and loss analyses confirmed that the focal loss model attended to intrinsic morphological features rather than object size. This study provides a practical, preprocessing-free approach for reliable herbal-medicine authentication in field conditions.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Smart Farm Science, Kyung Hee University, Yongin 17104, Republic of Korea.
- Interdisciplinary Program in IT-Bio Convergence System, Kyung Hee University, Yongin 17104, Republic of Korea.
ORCID-Profile: Yea-Jin Park, Dae-Hyun Jung
Zitieren
APA
Park, Y. J., & Jung, D. H. (2026). CNN-Based Classification of Ziziphus Seeds with Focal Loss for Overcoming Size-Based Shortcut Learning. Biosensors, 16(6), 320. https://doi.org/10.3390/bios16060320
Vancouver
Park YJ, Jung DH. CNN-Based Classification of Ziziphus Seeds with Focal Loss for Overcoming Size-Based Shortcut Learning. Biosensors. 2026;16(6):320. doi:10.3390/bios16060320
BibTeX
@article{Park2026,
author = {Park, Yea-Jin and Jung, Dae-Hyun},
title = {CNN-Based Classification of Ziziphus Seeds with Focal Loss for Overcoming Size-Based Shortcut Learning},
journal = {Biosensors},
year = {2026},
volume = {16},
number = {6},
pages = {320},
doi = {10.3390/bios16060320}
}
EUVIMED