Proceedings of the
The Nineteenth International Conference on Computational Intelligence and Security (CIS 2023)
December 1 – 4, 2023, Haikou, China
Adaptive Fusion Learning on Kernel PLS Method with Derivative Preprocessing for FT-NIR Spectral Data
School of Mathematics and Statistics & Center for Data Analysis and Algorithm Technology, Guilin University of Technology, China.
ABSTRACT
Adaptive learning is a prevalent topic accompanied with the development of big data and intelligent computing in applications of rapid detection for some determinant issues in the eco-environmental or biological fields. In this paper, we study the improvement of the PLS regression method by the kernel transform tactic, and the theme of algorithmic fusion with derivative data preprocessing methods. Practically, the derivative-involved Savitzky-Golay smoother (SGS) and Whittaker smoother (WTK) methods are enhanced with adaptive tuning on flexible orders of derivative. Then these two methods are designed in fusion to the improved the kernel partial least squares (KPLS) model, for the optimization of Fourier transform near infrared (FT-NIR) analysis of soil samples, quantitatively. The experimental case validated that the proposed methodologies are well calibrated and predicted with appreciating modeling results. It is prospective to be extended for wide range applications. Keywords: FT-NIR data, Kernel PLS, SGS, WTK, Adaptive learning, Fusion optimization.
Download PDF
School of Mathematics and Statistics & Center for Data Analysis and Algorithm Technology, Guilin University of Technology, China.
ABSTRACT
Adaptive learning is a prevalent topic accompanied with the development of big data and intelligent computing in applications of rapid detection for some determinant issues in the eco-environmental or biological fields. In this paper, we study the improvement of the PLS regression method by the kernel transform tactic, and the theme of algorithmic fusion with derivative data preprocessing methods. Practically, the derivative-involved Savitzky-Golay smoother (SGS) and Whittaker smoother (WTK) methods are enhanced with adaptive tuning on flexible orders of derivative. Then these two methods are designed in fusion to the improved the kernel partial least squares (KPLS) model, for the optimization of Fourier transform near infrared (FT-NIR) analysis of soil samples, quantitatively. The experimental case validated that the proposed methodologies are well calibrated and predicted with appreciating modeling results. It is prospective to be extended for wide range applications. Keywords: FT-NIR data, Kernel PLS, SGS, WTK, Adaptive learning, Fusion optimization.
Download PDF