Calibration equations for the estimation of amino acid composition in whole soybeans were developed using partial least squares (PLS), artificial neural networks (ANN), and support vector machines (SVM) regression methods for five models of near-infrared (NIR) spectrometers. The effects of amino acid/protein correlation, calibration method, and type of spectrometer on predictive ability of the equations were analyzed. Validation of prediction models resulted in r 2 values from 0.04 (tryptophan) to 0.91 (leucine and lysine). Most of the models were usable for research purposes and sample screening. Concentrations of cysteine and tryptophan had no useful correlation with spectral information. Predictive ability of calibrations was dependent on the respective amino acid correlations to reference protein. Calibration samples with nontypical amino acid profiles relative to protein would be needed to overcome this limitation. The performance of PLS and SVM was significantly better than that of ANN. Choice of preferred modeling method was spectrometerdependent. Calibration equations for the estimation of amino acid composition in whole soybeans were developed using partial least squares (PLS), artificial neural networks (ANN), and support vector machines (SVM) regression methods for five models of near-infrared (NIR) spectrometers. The effects of amino acid/ protein correlation, calibration method, and type of spectrometer on predictive ability of the equations were analyzed. Validation of prediction models resulted in r 2 values from 0.04 (tryptophan) to 0.91 (leucine and lysine). Most of the models were usable for research purposes and sample screening.Concentrations of cysteine and tryptophan had no useful correlation with spectral information. Predictive ability of calibrations was dependent on the respective amino acid correlations to reference protein. Calibration samples with nontypical amino acid profiles relative to protein would be needed to overcome this limitation. The performance of PLS and SVM was significantly better than that of ANN. Choice of preferred modeling method was spectrometer-dependent.

The objectives of this research were: (1) to develop a technique for creating calibrations to predict the constituent concentrations of single maize kernels from near-infrared (NIR) hyperspectral image data, and (2) to evaluate the feasibility of an NIR hyperspectral imaging spectrometer as a tool for the quality analysis of single maize kernels. Single kernels of maize were analyzed by hyperspectral transmittance in the range of 750 to 1090 nm. The transmittance data were standardized using an opal glass transmission standard and converted to optical absorbance units. Partial least squares (PLS) regression and principal components regression (PCR) were used to develop predictive calibrations for moisture and oil content using the standardized absorbance spectra. Standard normal variate, detrending, multiplicative scatter correction, wavelength selection by genetic algorithm, and no preprocessing were compared for their effect on model predictive performance. The moisture calibration achieved a best standard error of cross-validation (SECV) of 1.20%, with relative performance determinant (RPD) of 2.74. The best oil calibration achieved an SECV of 1.38%, with an RPD of only 1.45. The performance and subsequent analysis of the oil calibration reveal the need for improved methods of single-seed reference analysis.

A key element of successful development of new soybean cultivars is availability of inexpensive and rapid methods for measurement of FA in seeds. Published research demonstrated applicability of NIR spectroscopy for FA profiling in oilseeds. The objectives of this study were to investigate the applicability of NIR spectroscopy for measurement of FA in whole soybeans and compare performance of calibration methods. Equations were developed using partial least squares (PLS), artificial neural networks (ANN), and support vector machines (SVM) regression methods. Validation results demonstrated that (i) equations for total saturates had the highest predictive ability (r 2 = 0.91-0.94) and were usable for quality assurance applications, (ii) palmitic acid models (r 2 = 0.80-0.84) were usable for certain research applications, and (iii) equations for stearic (r 2 = 0.49-0.68), oleic (r 2 = 0.76-0.81), linoleic (r 2 = 0.73-0.76), and linolenic (r 2 = 0.67-0.74) acids could be used for sample screening. The SVM models produced significantly more accurate predictions than those developed with PLS. ANN calibrations were not different from the other two methods. Reduction in the number of calibration samples reduced predictive ability of all equations. The rate of performance degradation of SVM models with sample reduction was the lowest. a r 2 is determination coefficient, SEP is SE of prediction corrected for bias, d is bias, RPD is relative predictive determinant, PLS is partial least squares, ANN is artifical neural networks, and LS-SVM is Least Squares support vector machines. b Model parameters provide number of latent variables for PLS, number of inputs and neurons in a hidden layer for ANN, and radial basis function bandwidth and complexity regularization parameter for LS-SVM. NIR MEASUREMENT OF SOYBEAN FATTY ACIDS 425 JAOCS, Vol. 83, no. 5 (2006) FIG. 2. Actual vs. predicted concentration plots for saturates and linolenic FA calibrations. Models were tested on sets of 180 (saturates) and 244 (linolenic) samples. The solid line on each plot represents the regression line.

No abstract

The use of near‐infrared spectroscopy (NIRS) for the prediction of whole‐grain triticale moisture and protein content was evaluated. Because triticale is genetically close to wheat, commercially available wheat prediction models for Foss Infratec analyzers were applied in a year‐by‐year basis to triticale samples harvested in Iowa between 2002 and 2006. Wheat models were not applicable to moisture prediction (SEPavg = 0.37% pt; expected SEP on wheat samples 0.15% pt), but usable for screening for protein (SEPavg = 0.38% pt; expected SEP on wheat samples 0.25% pt). Dedicated triticale calibrations were developed from 2002 to 2005 data. Prediction results for 2006 samples only were compared. Triticale calibrations performed better than wheat calibrations for 2006 samples (moisture SEPtriticale = 0.29% pt, SEPwheat = 0.50% pt; protein SEPtriticale = 0.30% pt, SEPwheat = 0.68% pt).