Multivariate Classification of Prunus Dulcis Varieties using Leaves of Nursery Plants and Near Infrared Spectroscopy

Near-Infrared data classification at phone terminal based on the combination of PCA and CS-RBFSVC algorithms

Near-infrared (NIR) spectroscopy is a non-destructive, efficient and convenient detection technology, with the emergence of portable NIR spectrometers, NIR mobile applications (APPs) come into being. The popularity of intelligent mobile phones provides an impetus to the research and development of NIR APPs, however, the primary functions such as operating the NIR spectrometers and collecting data cannot satisfy NIR users in the field of data processing. Herein, we propose an APP processing NIR data locally at the mobile terminal, by the comprehensive utilization of Principal Component Analysis (PCA) and Cuckoo Search algorithm optimized Support Vector Classifier with radial basis function (RBFSVC) kernel (CS-RBFSVC). 738 NIR samples of four drugs (Cydiodine Buccal Tablets, Sulfasalazine Enteric-coated Tablets, Dexamethasone Acetate Tablets, Vecuronium Bromide for Injection) were used as the validation objects to train and test the data classification model. Firstly, the original data were subjected to dimensional reduction through PCA for the purpose of compressing calculation amount. Secondly, the CS-RBFSVC model was utilized to classify the types of drugs and their manufacturers, moreover, the improved accuracy and efficiency by introducing Cuckoo Search (CS) algorithm into RBFSVC were proven in comparison with the conventional grid optimized RBFSVC (Grid-RBFSVC) and Linear Support Vector Classifier (Linear-SVC). Last but not least, an APP based on the proposed PCA and CS-RBFSVC model is developed and demonstrated to be able to classify the type of drugs with an accuracy of 100%, the accuracies of classifying the drugs' manufacturers were 100%, 100%, 98.3% and 90.7%, respectively. Conclusively, the proposed PCA and CS-RBFSVC based model can provide a low-consumption, high accuracy and quick strategy for NIR data classification and overcome the limitations of internal storage and operating speed at phone terminals, in conjunction with the portable NIR spectrometer, it is believed to push forward NIR technology into the instant detection and on-site inspection.

A Review of Machine Learning for Near-Infrared Spectroscopy

The analysis of infrared spectroscopy of substances is a non-invasive measurement technique that can be used in analytics. Although the main objective of this study is to provide a review of machine learning (ML) algorithms that have been reported for analyzing near-infrared (NIR) spectroscopy from traditional machine learning methods to deep network architectures, we also provide different NIR measurement modes, instruments, signal preprocessing methods, etc. Firstly, four different measurement modes available in NIR are reviewed, different types of NIR instruments are compared, and a summary of NIR data analysis methods is provided. Secondly, the public NIR spectroscopy datasets are briefly discussed, with links provided. Thirdly, the widely used data preprocessing and feature selection algorithms that have been reported for NIR spectroscopy are presented. Then, the majority of the traditional machine learning methods and deep network architectures that are commonly employed are covered. Finally, we conclude that developing the integration of a variety of machine learning algorithms in an efficient and lightweight manner is a significant future research direction.

Feasibility study on early identification of freshness decay of fresh-cut kiwifruit during cold chain storage by Fourier transform-near infrared spectroscopy combined with chemometrics

This work mainly aimed to evaluate the feasibility of Fourier transform-near infrared spectroscopy (FT-NIRS) combined with chemometrics in early identification of freshness decay of fresh-cut kiwifruit during simulated cold chain storage, with organoleptic evaluation as a reference. By linear discriminant analysis, the freshness decay could be identified after only 2 days of cold storage, corresponding to freshness level of 3.41 ± 0.27 N (hardness), 0.70 ± 0.05 g/kg (total acid), 8.62 ± 0.06 g/100 g (reducing sugars), 62.04 ± 1.03 mg/100 g (vitamin C) and 2.05 ± 0.11 log₁₀ CFU/g (total plate count). Organoleptic evaluators could not perceive the freshness decay that occurred after 2 days of cold storage. Moreover, the freshness decay could be well quantitatively predicted by partial least squares regression, with low RMSE_p (0.18-05.42) and high R² (0.90-0.96). FT-NIRS appears to be a promising option for early warning of the freshness decay of fresh-cut kiwifruit during cold chain storage, thereby preventing serious spoilage and ensuring fresh fruits for consumers. PRACTICAL APPLICATION: This work is based on the fact that fresh-cut kiwifruit is prone to freshness decay under unstable cold chain conditions, using FT-NIRS combined with chemometrics to identify the freshness decay early and rapidly, to a certain extent, early warn freshness decay and effectively prevent serious spoilage. The technology can be used for food regulatory agencies to monitor the freshness of fresh-cut kiwifruit, and can also be applied for fruit processing enterprises and dealers to ensure the freshness and high quality of fresh-cut kiwifruit.

Geographic Origin Discrimination of Millet Using Vis-NIR Spectroscopy Combined with Machine Learning Techniques

Millet is a primary food for people living in the dry and semi-dry regions and is dispersed within most parts of Europe, Africa, and Asian countries. As part of the European Union (EU) efforts to establish food originality, there is a global need to create Protected Geographical Indication (PGI) and Protected Designation of Origin (PDO) of crops and agricultural products to ensure the integrity of the food supply. In the present work, Visible and Near-Infrared Spectroscopy (Vis-NIR) combined with machine learning techniques was used to discriminate 16 millet varieties (n = 480) originating from various regions of China. Five different machine learning algorithms, namely, K-nearest neighbor (K-NN), Linear discriminant analysis (LDA), Logistic regression (LR), Random Forest (RF), and Support vector machine (SVM), were used to train the NIR spectra of these millet samples and to assess their discrimination performance. Visible cluster trends were obtained from the Principal Component Analysis (PCA) of the spectral data. Cross-validation was used to optimize the performance of the models. Overall, the F-Score values were as follows: SVM with 99.5%, accompanied by RF with 99.5%, LDA with 99.5%, K-NN with 99.1%, and LR with 98.8%. Both the linear and non-linear algorithms yielded positive results, but the non-linear models appear slightly better. The study revealed that applying Vis-NIR spectroscopy assisted by machine learning technique can be an essential tool for tracing the origins of millet, contributing to a safe authentication method in a quick, relatively cheap, and non-destructive way.

Identification of Amaranthus Species Using Visible-Near-Infrared (Vis-NIR) Spectroscopy and Machine Learning Methods

The feasibility of rapid and non-destructive classification of six different Amaranthus species was investigated using visible-near-infrared (Vis-NIR) spectra coupled with chemometric approaches. The focus of this research would be to use a handheld spectrometer in the field to classify six Amaranthus sp. in different geographical regions of South Korea. Spectra were obtained from the adaxial side of the leaves at 1.5 nm intervals in the Vis-NIR spectral range between 400 and 1075 nm. The obtained spectra were assessed with four different preprocessing methods in order to detect the optimum preprocessing method with high classification accuracy. Preprocessed spectra of six Amaranthus sp. were used as input for the machine learning-based chemometric analysis. All the classification results were validated using cross-validation to produce robust estimates of classification accuracies. The different combinations of preprocessing and modeling were shown to have a classification accuracy of between 71% and 99.7% after the cross-validation. The combination of Savitzky-Golay preprocessing and Support vector machine showed a maximum mean classification accuracy of 99.7% for the discrimination of Amaranthus sp. Considering the high number of spectra involved in this study, the growth stage of the plants, varying measurement locations, and the scanning position of leaves on the plant are all important. We conclude that Vis-NIR spectroscopy, in combination with appropriate preprocessing and machine learning methods, may be used in the field to effectively classify Amaranthus sp. for the effective management of the weedy species and/or for monitoring their food applications.

Estimation of Andrographolides and Gradation of Andrographis paniculata Leaves Using Near Infrared Spectroscopy Together With Support Vector Machine

Andrographis paniculata (Burm. F) Nees, has been widely used for upper respiratory tract and several other diseases and general immunity for a historically long time in countries like India, China, Thailand, Japan, and Malaysia. The vegetative productivity and quality with respect to pharmaceutical properties of Andrographis paniculata varies considerably across production, ecologies, and genotypes. Thus, a field deployable instrument, which can quickly assess the quality of the plant material with minimal processing, would be of great use to the medicinal plant industry by reducing waste, and quality grading and assurance. In this paper, the potential of near infrared reflectance spectroscopy (NIR) was to estimate the major group active molecules, the andrographolides in Andrographis paniculata, from dried leaf samples and leaf methanol extracts and grade the plant samples from different sources. The calibration model was developed first on the NIR spectra obtained from the methanol extracts of the samples as a proof of concept and then the raw ground samples were estimated for gradation. To grade the samples into three classes: good, medium and poor, a model based on a machine learning algorithm - support vector machine (SVM) on NIR spectra was built. The tenfold classification results of the model had an accuracy of 83% using standard normal variate (SNV) preprocessing.