LSSVM Models Research Articles

Non-destructive determination of total polyphenols content and classification of storage periods of Iron Buddha tea using multispectral imaging system

Total polyphenols is a primary quality indicator in tea which is consumed worldwide. The feasibility of using near infrared reflectance (NIR) spectroscopy (800–2500nm) and multispectral imaging (MSI) system (405–970nm) for prediction of total polyphenols contents (TPC) of Iron Buddha tea was investigated in this study. The results revealed that the predictive model by MSI using partial least squares (PLS) analysis for tea leaves was considered to be the best in non-destructive and rapid determination of TPC. Besides, the ability of MSI to classify tea leaves based on storage period (year of 2004, 2007, 2011, 2012 and 2013) was tested and the classification accuracies of 95.0% and 97.5% were achieved using LS-SVM and BPNN models, respectively. These overall results suggested that MSI together with suitable analysis model is a promising technology for rapid and non-destructive determination of TPC and classification of storage periods in tea leaves.

Measurement of soil organic matter and available K based on SPA-LS-SVM

Visible and short wave infrared spectroscopy (Vis/SW-NIRS) was investigated in the present study for measurement of soil organic matter (OM) and available potassium (K). Four types of pretreatments including smoothing, SNV, MSC and SG smoothing+first derivative were adopted to eliminate the system noises and external disturbances. Then partial least squares regression (PLSR) and least squares-support vector machine (LS-SVM) models were implemented for calibration models. The LS-SVM model was built by using characteristic wavelength based on successive projections algorithm (SPA). Simultaneously, the performance of LSSVM models was compared with PLSR models. The results indicated that LS-SVM models using characteristic wavelength as inputs based on SPA outperformed PLSR models. The optimal SPA-LS-SVM models were achieved, and the correlation coefficient (r), and RMSEP were 0. 860 2 and 2. 98 for OM and 0. 730 5 and 15. 78 for K, respectively. The results indicated that visible and short wave near infrared spectroscopy (Vis/SW-NIRS) (325 approximately 1 075 nm) combined with LS-SVM based on SPA could be utilized as a precision method for the determination of soil properties.

Predicting soluble solid content in intact jaboticaba [Myrciaria jaboticaba (Vell.) O. Berg] fruit using near-infrared spectroscopy and chemometrics

The aim of this study was to evaluate the potential of near-infrared reflectance spectroscopy (NIR) as a rapid and non-destructive method to determine soluble solid content (SSC) in intact jaboticaba [Myrciaria jaboticaba (Vell.) O. Berg] fruit. Multivariate calibration techniques were compared with pre-processed data and variable selection algorithms, such as partial least squares (PLS), interval partial least squares (iPLS), a genetic algorithm (GA), a successive projections algorithm (SPA) and nonlinear techniques (BP-ANN, back propagation of artificial neural networks; LS-SVM, least squares support vector machine) were applied to building the calibration models. The PLS model produced prediction accuracy (R2=0.71, RMSEP=1.33°Brix, and RPD=1.65) while the BP-ANN model (R2=0.68, RMSEM=1.20°Brix, and RPD=1.83) and LS-SVM models achieved lower performance metrics (R2=0.44, RMSEP=1.89°Brix, and RPD=1.16). This study was the first attempt to use NIR spectroscopy as a non-destructive method to determine SSC jaboticaba fruit.

Non-destructive measurement of crude fat and crude protein contents of oats from different geographical zones in China using visible and near-infrared spectroscopy

respectively. The correct answer rate of the LS-SVM model was 98.3%, better than that of SIMCA (75.0%). LS-SVM models were also established for the content determination of crude fat and crude protein of oats. Eleven and thirteen wavelength variables were selected by successive projections algorithm for crude protein and crude fat analyses, respectively. The correlation coefficients were 0.814 and 0.915, and the root mean square errors of validation were 0.338 and 0.389 for the crude protein and crude fat analyses, respectively. Overall Vis-NIRS is a promising technique for the fast and reliable determination of geographical origins and crude fat content of oats. The determination of crude protein content needs to be further researched.

Introducing evolving Takagi–Sugeno method based on local least squares support vector machine models

In this study, an efficient local online identification method based on the evolving Takagi–Sugeno least square support vector machine (eTS-LS-SVM) for nonlinear time series prediction is introduced. As an innovation, this paper has applied the nonlinear models, i.e. local LS-SVM models, as the consequence parts of the fuzzy rules, instead of the linear models used in the conventional evolving TS fuzzy models. In each step, the proposed learning approach includes two phases. The fuzzy rules (rule premise) are first created and updated adaptively based on a sequential clustering technique to obtain the structure of TS model. Then, the parameters of each local LS-SVM model (rule consequence) are recursively updated by deriving a new recursive algorithm (a local decremental and incremental procedure) to minimize the local modelling error and trace the process’s dynamics. Besides, a new learning algorithm based on the recursive gradient-based method is used to adaptively update the meta-parameters of the LS-SVM models. Comparison of the suggested method with some of the previous approaches based on the online prediction of the nonlinear time series has shown that the introduced identification algorithm has a proper performance in terms of learning and generalization abilities while having a lower redundancy.

Application of visible and near infrared spectroscopy for rapid and non-invasive quantification of common adulterants in Spirulina powder

The authentication of food products is critically important in a global economy in public-health and economic terms. The specific aims of this study were to evaluate the application of full-spectrum and NIR spectroscopy and to evaluate the adoption of PLS and LS-SVM models to accomplish a rapid and non-invasive quantification of the two common adulterants, flour and mungbean powder, in Spirulina powder. The results showed that, using all treatment sets, only the LS-SVM models were adequate in predicting either adulterant under both full spectra and NIR spectra. The use of NIR spectra would allow detection of adulterants even when masked by food dyes. Design value analysis indicated that the benefits per unit cost of applying the NIR spectra to quantify adulterants in Spirulina powder significantly exceeded that of using full spectra, and that the value of employing the LS-SVM models under NIR spectra exceeded that of using the PLS models.

Classification of Raw and RoastedSemen CassiaeSamples with the Use of Fourier Transform Infrared Fingerprints and Least Squares Support Vector Machines

Raw and roasted Semen Cassiae seeds, a complex traditional Chinese medicine (TCM), are used as examples to research and develop a method of classification analysis based on measurements of Fourier transform infrared (FT-IR) spectral fingerprints. Eighty samples of the TCM were measured in the mid-infrared range, 400-2000 cm(-1) (KBr pellets), and the complex overlapping spectra were submitted for interpretation to a principal component analysis least squares support vector machine (PC-LS-SVM), kernel principal component analysis least squares support vector machine (KPC-LS-SVM), and radial basis function artificial neural networks (RBF-ANN). The LS-SVM models were developed with an RBF kernel function and a grid search technique. Training models were constructed with the use of raw and first-derivative spectra and these were then verified by another data set containing both raw and roasted spectral objects. It was demonstrated that the first-derivative data set produced the best separation of the spectral objects. In general, satisfactory analytical performance was obtained with the PC-LS-SVM, KPC-LS-SVM, and RBF-ANN training models and with the classification of the verification spectral objects. With regard to chemometrics modeling, the performance of KPC-LS-SVM was somewhat more economical than that of the PC-LS-SVM model. It would appear that the latter relatively simple model would be sufficient for application to most small to medium sized FT-IR fingerprint data sets, but with larger matrices the more complex models, such as the RBF-ANN and KPC-LS-SVM, may be more advantageous on a computational basis.