Spectral Wavelength Division for Multiview Deep Calibration Modeling
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 9800810-9800810
DOI: 10.1109/tim.2026.3697064 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Junhua Zheng und 4 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Zhejiang Provincial Natural Science Foundation of China, Jiangsu Provincial Scientific Research Center of Applied Mathematics.
Abstract (Originalsprache)
In most chemometrics modeling and application scenarios, we often deal with flat data matrices, which means the modeling dataset often has very high-dimensional spectra variables and a small number of training data samples. From the viewpoint of machine learning, this has a great potential risk to cause an overfitting problem, which may largely influence the calibration performance. In this article, a multiview deep learning framework is proposed for calibration modeling. By dividing the whole spectral wavelength into different parts, several subview deep learning models are constructed for more refined calibrations in specific areas. Then, an information fusion strategy is proposed for ensemble calibration results from different subviews. Based on this specific integration of a deep, layerwise residual principal component regression (PCR) architecture within a wavelength-partitioned multiview framework, the overfitting risk can be reduced, and the calibration performance can be significantly boosted. Besides, a possible piecewise linearization can be obtained through the wavelength channel for nonlinear spectra data analysis. Detailed comparative studies are carried out between shallow and deep models under the multiview modeling framework, based on which the superiority of the multiview deep learning model is confirmed.
Förderung
- National Natural Science Foundation of China
- Zhejiang Provincial Natural Science Foundation of China
- Jiangsu Provincial Scientific Research Center of Applied Mathematics
Zitieren
APA
Zheng, J., Zhou, H., Lyu, Y., Yang, Z., & Ge, Z. (2026). Spectral Wavelength Division for Multiview Deep Calibration Modeling. IEEE Transactions on Instrumentation and Measurement, 75, 9800810-9800810. https://doi.org/10.1109/tim.2026.3697064
Vancouver
Zheng J, Zhou H, Lyu Y, Yang Z, Ge Z. Spectral Wavelength Division for Multiview Deep Calibration Modeling. IEEE Transactions on Instrumentation and Measurement. 2026;75:9800810-9800810. doi:10.1109/tim.2026.3697064
BibTeX
@article{Zheng2026,
author = {Zheng, Junhua and Zhou, Hansong and Lyu, Yuting and Yang, Zeyu and Ge, Zhiqiang},
title = {Spectral Wavelength Division for Multiview Deep Calibration Modeling},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {9800810--9800810},
doi = {10.1109/tim.2026.3697064}
}
EUVIMED