Related Experiment Video
Updated: May 10, 2026

Synthesis and Operation of Fluorescent-core Microcavities for Refractometric Sensing
Published on: March 13, 2013
Weighted fusion of multiple models for wavelength selection
Tianling Zeng1, Zhiyu Wen, Zhongquan Wen
1Key Laboratory of Fundamental Science on Micro/Nano-Device and System Technology, Chongqing University, Chongqing 400030, China.
A novel weighted fusion method improves wavelength selection for spectral data analysis. This approach enhances the accuracy and stability of partial least squares (PLS) models, offering a more efficient alternative for practical applications.
Area of Science:
- Chemometrics
- Spectroscopy
- Data Analysis
Background:
- Multivariate calibration of spectral data often faces challenges with selecting informative wavelengths.
- Existing methods like genetic algorithms and uninformative variable elimination have limitations in accuracy and efficiency.
- Partial least squares (PLS) modeling is widely used but can be improved by optimized wavelength selection.
Purpose of the Study:
- To introduce a new weighted fusion method for wavelength selection in multivariate calibration.
- To enhance the accuracy and stability of wavelength selection compared to existing methods.
- To improve the predictive performance of partial least squares (PLS) models.
Main Methods:
- A weighted fusion approach combining regression coefficients from multiple models.
- Weighting is determined by minimizing the mean square error.
- Application to three near-infrared spectral datasets for partial least squares (PLS) modeling.
Main Results:
- The proposed weighted fusion method effectively identifies informative wavelengths.
- It significantly enhances the prediction accuracy and stability of PLS models.
- Performance was validated against full-spectrum PLS, genetic algorithm-based PLS, and uninformative variable elimination-based PLS.
Conclusions:
- The weighted fusion method offers superior wavelength selection for spectral data.
- It provides a simpler, more efficient, and highly effective alternative for practical chemometric applications.
- This method improves the predictive capabilities of multivariate calibration models.
Related Concept Videos
UV–Vis Spectroscopy: Woodward–Fieser Rules
Ultraviolet and Visible (UV–Vis) Spectroscopy: Overview
Frequency-dependent Selection
IR Spectrum Peak Splitting: Symmetric vs Asymmetric Vibrations
Wave Parameters
Pharmacokinetic Models: Comparison and Selection Criterion
Physiological models take a detailed approach by considering specific molecular processes. They can predict drug distribution, metabolism, and elimination changes, providing a comprehensive understanding of how drugs interact with the body.