Artificial neural networks in classification of NIR spectral data: Design of the training set

Wu, Wen; Walczak, Beata; Massart, D.L.; Heuerding, S.; Erni, F.; Prebble, K.A.

doi:10.1016/0169-7439(95)00077-1

Cited by 214 publications

(121 citation statements)

References 21 publications

Supporting

Mentioning

120

Contrasting

Unclassified

Order By: Relevance

“…These networks have been little used for the classification of infrared spectra and have never been used directly as an interpretation system for general compounds. They have been used only for feasibility studies [13], selection of the training set for another interpretation system [14][15][16], or specific studies [17][18]. After a short introduction to Kohonen networks and its learning process, we present results on the clustering of infrared spectra of organic compounds from a commercial database [19].…”

Section: Introductionmentioning

confidence: 99%

Clustering of infrared spectra with Kohonen networks

Cleva

Cachet²,

Cabrol‐Bass³

1999

Analusis

View full text Add to dashboard Cite

Abstract. The design of systems for spectral data interpretation requires clustering of chemical compounds based on their spectral characteristics. Kohonen networks have been shown to be efficient tools to achieve this clustering. These auto-organising systems perform a mapping between a high-dimensional variable space and a two-dimensional one. An application to infrared spectra of organic compounds is presented here. The non-supervised learning algorithm used allows classification of compounds by spectral characteristics without a priori knowledge. An analysis of the distribution of spectra on the resulting maps is used to build models for predicting the presence or absence of specific structural features. The performance of the models in recognising structural features is discussed and compared with the prediction of a multilayered feed forward network (MLFFN). Localisation of compounds wrongly classified by the MLFFN on the Kohonen maps allows to establish a link between the supervised and the unsupervised approaches.

show abstract

Section: Introductionmentioning

confidence: 99%

Clustering of infrared spectra with Kohonen networks

Cleva

Cachet²,

Cabrol‐Bass³

1999

Analusis

View full text Add to dashboard Cite

show abstract

“…In this study, 5 kinds of sample set partition methods, such as sample set partitioning based on joint X-Y distances (SPXY) [12], Kennard-Stone(KS) [13], random sampling (RS), duplex, CG, were compared and discussed.…”

Section: Data Processing Methodsmentioning

confidence: 99%

Determination of Total Volatile Basic Nitrogen (TVB-N) Content in Beef by Hyperspectral Imaging Technique

Liu

2016

MATEC Web of Conferences

View full text Add to dashboard Cite

Abstract. Non-destructive determination of TVB-N content in beef using hyperspectral imaging (HSI) technique was evaluated. In order to create a robust model to predict the TVB-N content in beef, partition of sample set, spectral pretreatment, and the optimum wavelength selection were discussed. After the beef sample set was parted by concentration gradient (CG) algortithm, and the spectra of beef samples were preprocessed by standard normalized variate (SNV) combined with auto scale(AS), the partial least square regression (PLSR) model was established using the full spectral range, which had the best prediction abilities with R cv 2 of 0.9124, R p 2 of 0.8816, RMSECV of 1.5889, and RMSEP of 1.7719, respectively. After the optimum wavelengths which is closely related to the TVB-N content of beef samples was obtained using the competitive adaptive reweighted (CARS) algorithm, a new PLSR model was established using the optimum wavelengths, which had outstanding prediction abilities with R cv 2 of 0.9235, R p 2 of 0.9241, RMSECV of 1.4881, and RMSEP of 1.4882, respectively.The study showed that HSI is a powerful technique to predict the TVB-N content in beef by a nondestructive way.

show abstract

“…The advantages of this algorithm are that the calibration samples always map the measured region of the input variable space completely with respect to the induced metric and that the no validation samples fall outside the measured region. The Kennard and Stones algorithm has been considered one of the best ways to build training and validation sets [28,29]. Using Kennard and Stones algorithm, the entire dataset was divided into two subsets: a training set of 71 compounds, and a validation set including the remaining 30 compounds.…”

Section: Datasetmentioning

confidence: 99%

A simple 2D-QSPR model for the prediction of Setschenow constants of organic compounds

Fan

2016

Maced. J. Chem. Chem. Eng.

View full text Add to dashboard Cite

A quantitative structure-property relationship (QSPR) analysis of the Setschenow constants (K salt ) of organic compounds in a sodium chloride solution was carried out using only two-dimensional (2D) descriptors as input parameters. The whole set of 101 compounds was split into a training set of 71 compounds and a validation set of 30 compounds by means of the Kennard and Stones algorithm. A general four-parameter equation, with correlation coefficient (R) of 0.887 and standard error of estimation (s) of 0.031, was obtained by stepwise multilinear regression analysis (MLRA) on the training set. The reliability and robustness of the present model was verified with leave-one-out cross-validation, randomization tests, and the external validation set. All of the descriptors contained in this model are calculated directly from the molecular 2D structures; thus, this model can be used to easily predict the K salt of other compounds not involved in the present dataset.Keywords: QSPR; Setschenow constants; 2D descriptor; multilinear regression analysis ЕДНОСТАВЕН 2D-QSPR-МОДЕЛ ЗА ПРЕДВИДУВАЊЕ НА КОНСТАНТИТЕ НА SETSCHENOW ЗА ОРГАНСКИТЕ СОЕДИНЕНИЈААнализата на квантитативната зависност на структурата и својствата (QSPR) на константите на Setschenow (K salt ) на органските соединенија во раствор од натриум хлорид е извршена користејќи само дводимензионални (2D) дескриптори како влезни параметри. Целото множество од 101 соединение беше поделено во множество за подготовка од 71 соединение и множество за валидација од 30 соединенија според алгоритамот на Kennard и Stones. Од множеството за подготовка со мултилинеарна регресиона анализа MLRA е добиена општа четирипараметарска равенка со коефициент на корелација R = 0,887 и стандардна грешка на процената s = 0,031. Веродостојноста и робусноста на овој модел беше верифицирана со повеќе тестови: вкрстена валидација со испуштање на еден параметар, тест на рандомизација, како и екстерно множество за валидација. Сите дескриптори содржат модел што се пресметува директно од молекулските 2D-структури, и така овој модел може да се користи за едноставно предвидување на K salt на други соединенија што не се вклучени во ова множество на податоци.

show abstract

Artificial neural networks in classification of NIR spectral data: Design of the training set

Cited by 214 publications

References 21 publications

Clustering of infrared spectra with Kohonen networks

Clustering of infrared spectra with Kohonen networks

Determination of Total Volatile Basic Nitrogen (TVB-N) Content in Beef by Hyperspectral Imaging Technique

A simple 2D-QSPR model for the prediction of Setschenow constants of organic compounds

Contact Info

Product

Resources

About