Local tangent space alignment and relevance vector machine as nonlinear methods for estimating sensory quality of tea using NIR spectroscopy

Machine Learning in Chemical Product Engineering: The State of the Art and a Guide for Newcomers

“…Chemometrics methods, such as ANOVA or PCA, are most often applied before ML methods to select and compress the original data [156][157][158][159][160][161][162][163][164]. Feature extraction methods such as Fourier analysis or Si-PLS are also used for extracting relevant information or optimal spectral interval from high dimensional spectra measurements [156,158].…”

“…This step is particularly important when working with spectral data as it prevents many irrelevant or redundant spectrum variables from being introduced and, therefore, decreases the complexity and size of the variable space and improves the precision of the model [164]. Supervised ML methods, such as ANN, SVM and RF, are then applied to link the sensory properties with the process parameters, the ingredients or the microstructure of the tested products.…”

“…Great effort in sensory evaluation has been made in an attempt to replace the subjective sensory evaluation with the use of more objective, robust and reproducible instrumental and analytical measurements, in view of overcoming the associated uncertainty, imprecision and time demand of the classical sensory evaluation procedures [155,164]. Analytical methods, such as near infrared (NIR), Fourier transform infrared (FT-IR) and Raman spectroscopy analyses, or other instrumental methods combined with ML or chemometrics algorithms, have been used as efficient methods to quickly evaluate the biochemical or physical characteristics of food products and decode their correlations to sensory attributes.…”

“…The authors in [164] compared several linear and nonlinear dimensionality reduction (kernel PCA, sparse PCA, local tangent space alignment, PCA and multidimensional scaling) and regression methods (relevance vector machine, back-propagation ANN and PLS) for estimating the sensory quality of tea from NIR spectra. In this application, nonlinear methods displayed better performance due to existing non linearity in tea components and NIR spectra.…”

Determination of polyphenol antioxidants in the samples of green tea. The characteristic chromatographic profiles

“…В последние годы все более широкое распространение в аналитической химии находят многомерные методы обработки данных [16][17][18][19][20][21][22], что помогает выявить скрытые закономерности, отделить полезный сигнал от шума, а также представить данные в более удобной для интерпретации и визуализации форме. Одним из таких методов является метод главных компонент (МГК), заключающийся в построении ортогональных линий, направление которых совпадает с максимальным увеличением дисперсии в изучаемой системе.…”