[Show abstract][Hide abstract] ABSTRACT: Traditional ensemble regression algorithms such as BAgging Partial Least Squares (BAPLS) and BOosting Partial Least Squares (BOPLS) do not perform very well in the data set that is relatively small or contaminated by random noise. To make the method robust and improve its prediction ability, inspired from bias-variance-covariance decomposition, we propose an improved ensemble partial least squares method based on the diversity. The new method is applied to quantitative analysis of Near InfraRed (NIR) data sets. A comparative study between the proposed method and other previous methods including BAPLS and BOPLS on two NIR data sets is provided. Experimental results show that the proposed method can achieve better performance than other methods.
[Show abstract][Hide abstract] ABSTRACT: In order to eliminate the lower order polynomial interferences, a new quantitative calibration algorithm "Baseline Correction Combined Partial Least Squares (BCC-PLS)", which combines baseline correction and conventional PLS, is proposed. By embedding baseline correction constraints into PLS weights selection, the proposed calibration algorithm overcomes the uncertainty in baseline correction and can meet the requirement of on-line attenuated total reflectance Fourier transform infrared (ATR-FTIR) quantitative analysis. The effectiveness of the algorithm is evaluated by the analysis of glucose and marzipan ATR-FTIR spectra. BCC-PLS algorithm shows improved prediction performance over PLS. The root mean square error of cross-validation (RMSECV) on marzipan spectra for the prediction of the moisture is found to be 0.53%, w/w (range 7-19%). The sugar content is predicted with a RMSECV of 2.04%, w/w (range 33-68%).
[Show abstract][Hide abstract] ABSTRACT: In this paper, based on asymmetric least squares smoothing, a new algorithm for multiple spectra baseline correction is proposed. By means of the similarity among the multiple spectra, the algorithm estimates the baselines by penalizing the differences in the baseline corrected signals, which makes the algorithm possible to eliminate scatter effects on the spectra. In addition, a relaxation factor which measures the similarity of the baseline corrected spectra is incorporated into the optimization model and an alternate iteration strategy is used to solve the optimization problem. The proposed algorithm is fast and can output multiple baselines simultaneously. Experimental results on both simulated data and real data demonstrate the effectiveness and efficiency of the algorithm.
[Show abstract][Hide abstract] ABSTRACT: FT-IR and two-dimensional correlation spectroscopy (2D-IR) technology were applied to discriminate Chinese Sauce liquor from different fermentation positions (top, middle and bottom of fermentation cellar) for the first time. The liquors at top, middle and bottom of fermentation cellar, possessed the characteristic peaks at 1731 cm−1, 1733 cm−1 and 1602 cm−1, respectively. In the 2D correlation infrared spectra, the differences were amplified. A strong auto-peak at 1725 cm−1 showed in the 2D spectra of the Top Liquor, which indicated that the liquor might contain some ester compounds. Different from Top Liquor, three auto-peaks at 1695, 1590 and 1480 cm−1 were identified in 2D spectra of Middle Liquor, which were the characteristic absorption of acid, lactate. In 2D spectra of Bottom Liquor, two auto-peaks at 1570 and 1485 cm−1 indicated that lactate was the major component. As a result, FT-IR and 2D-IR correlation spectra technology provided a rapid and effective method for the quality analysis of the Sauce liquor.