EUVIMEDEUVIMED entdecken
Crossref

From Memorization to Generalization: L1‐BO‐SVR for Robust Vis‐NIR Prediction of Mango Dry Matter Across Diverse Cultivars, Origins, and Harvest Years

Qing-Xiao Ma, Hao-Xuan Kan, Rui Chen, Wen-Fei Zhang, Guo-Ao Xie, Hao Liu

Journal of Food Science · 2026-11 · 91 (11)

DOI: 10.1111/1750-3841.71616 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Food Science, online erschienen am 05.11.2026. Verfasst von Qing-Xiao Ma und 5 weiteren Autor:innen. Beteiligte Einrichtung: Faculty of Electronic Information Engineering, Huai'an University, Huaian, China. Hauptthemen laut MEDLINE: Mangifera. Die Arbeit nennt 32 Literaturangaben. Gefördert von: Production-Research-Development (R&D) projects of Jiangsu Province, Jiangsu Engineering Research Center of Lake Environment Remote Sensing Technologies.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Mango ( Mangifera indica L.) dry matter (DM) content is a critical quality indicator determining consumer acceptance and commercial value, yet conventional oven‐drying methods are destructive and time‐consuming. Visible and near‐infrared (Vis‐NIR) spectroscopy offers a rapid and non‐consumptive alternative, but existing models suffer from poor generalizability due to spectrally homogeneous training sets. To address this limitation, we designed a diverse sampling strategy encompassing three cultivars, two production regions, and three harvest years, yielding 2,787 mango samples with DM content ranging from 9.46% to 22.95%. We propose a hybrid modeling framework named L1‐BO‐SVR integrating L1‐regularized feature selection with Bayesian optimization (BO) enhanced support vector regression (SVR). The L1‐BO‐SVR model selected 94 informative wavelengths from 234 variables and achieved excellent predictive performance, with R 2 P of 0.947 and RMSE P of 0.443%, substantially outperforming full‐spectrum models with R 2 P of 0.895 and conventional preprocessing‐model combinations. Critically, the small performance gaps among training, cross‐validation, and test sets confirm robust generalization with minimal overfitting. These results demonstrate that diverse sampling combined with L1‐BO‐SVR effectively overcomes the “machine memorization” pitfall, enabling reliable and generalizable Vis‐NIR prediction of mango DM content across diverse production conditions, a critical advantage for practical deployment in real‐world supply chains.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

MangiferaSpectroscopy, Near-InfraredSupport Vector MachineBayes TheoremFruit

Stichwörter der Autor:innen

Vis‐NIR spectroscopydry matter contentmachine learningmangomodel generalizability

Autor:innen und Institutionen

  • Faculty of Electronic Information Engineering, Huai'an University, Huaian, China.
  • Faculty of Delaware, Huai'an University, Huaian, China.
  • Institute of Energy, Hefei Comprehensive National Science Center, Hefei, China.

Förderung

  • Production-Research-Development (R&D) projects of Jiangsu Province
  • Jiangsu Engineering Research Center of Lake Environment Remote Sensing Technologies

Zitieren

APA

Ma, Q. X., Kan, H. X., Chen, R., Zhang, W. F., Xie, G. A., & Liu, H. (2026). From Memorization to Generalization: L1‐BO‐SVR for Robust Vis‐NIR Prediction of Mango Dry Matter Across Diverse Cultivars, Origins, and Harvest Years. Journal of Food Science, 91(11). https://doi.org/10.1111/1750-3841.71616

Vancouver

Ma QX, Kan HX, Chen R, Zhang WF, Xie GA, Liu H. From Memorization to Generalization: L1‐BO‐SVR for Robust Vis‐NIR Prediction of Mango Dry Matter Across Diverse Cultivars, Origins, and Harvest Years. Journal of Food Science. 2026;91(11). doi:10.1111/1750-3841.71616

BibTeX
@article{Ma2026,
  author = {Ma, Qing-Xiao and Kan, Hao-Xuan and Chen, Rui and Zhang, Wen-Fei and Xie, Guo-Ao and Liu, Hao},
  title = {From Memorization to Generalization: L1‐BO‐SVR for Robust Vis‐NIR Prediction of Mango Dry Matter Across Diverse Cultivars, Origins, and Harvest Years},
  journal = {Journal of Food Science},
  year = {2026},
  volume = {91},
  number = {11},
  doi = {10.1111/1750-3841.71616}
}
Kostenlos starten →