Improvement of spectral calibration for food analysis through multi-model fusion

Rapid determination of hemoglobin concentration by a novel ensemble extreme learning machine method combined with near-infrared spectroscopy

A novel ensemble extreme learning machine (ELM) approach that combines Monte Carlo (MC) sampling and least absolute shrinkage and selection operator (LASSO), named as MC-LASSO-ELM, is proposed to determine hemoglobin concentration of blood. It employs MC sampling to randomly select samples from the training set and LASSO further to choose variables from selected samples to establish plenty of ELM sub-models. The final prediction is obtained by combining the predictions of these sub-models. Combined with near-infrared spectroscopy, MC-LASSO-ELM is used to determine the hemoglobin concentration of blood. Compared with ELM, MC-ELM and LASSO-ELM, MC-LASSO-ELM can obtain the best stability and highest accuracy.

A new ensemble modeling method for multivariate calibration of near infrared spectra

Ensemble modeling has gained increasing attention for improving the performance of quantitative models in near infrared (NIR) spectral analysis. Based on Monte Carlo (MC) resampling, least absolute shrinkage and selection operator (LASSO) and partial least squares (PLS), a new ensemble strategy named MC-LASSO-PLS is proposed for NIR spectral multivariate calibration. In this method, the training subsets for building the sub-models are generated by sampling from both samples and variables to ensure the diversity of the models. In detail, a certain number of samples as sample subsets are randomly selected from training set. Then, LASSO is used to shrink the variables of the sample subset to form the training subset, which is used to build the PLS sub-model. This process is repeated N times and N sub-models are obtained. Finally, the predictions of these sub-models are used to produce the final prediction by simple average. The prediction ability of the proposed method was compared with those of LASSO-PLS, MC-PLS and PLS models on the NIR spectra of corn, blend oil and orange juice samples. The superiority of MC-LASSO-PLS in prediction ability is demonstrated.

Ensemble of extreme learning machines for multivariate calibration of near-infrared spectroscopy

Inspired by the attractive features of extreme learning machine (ELM), a simple ensemble ELM algorithm, named EELM, is proposed for multivariate calibration of near-infrared spectroscopy. Such an algorithm takes full advantage of random initialization of the weights of the hidden layer in ELM for obtaining the diversity between member models. Also, by combining a large number of member models, the stability of the final prediction can be greatly improved and the ensemble model outperforms its best member model. Compared with partial least-squares (PLS), the superiority of EELM is attributed to its inherent characteristics of high learning speed, simple structure and excellent predictive performance. Three NIR spectral datasets concerning solid samples are used to verify the proposed algorithm in terms of both the accuracy and robustness. The results confirmed the superiority of EELM to classic PLS. Also, even if the experiment is done on NIR datasets, it provides a good reference for other spectral calibration.

Non-Destructive Spectroscopic Techniques and Multivariate Analysis for Assessment of Fat Quality in Pork and Pork Products: A Review

Fat is one of the most important traits determining the quality of pork. The composition of the fat greatly influences the quality of pork and its processed products, and contribute to defining the overall carcass value. However, establishing an efficient method for assessing fat quality parameters such as fatty acid composition, solid fat content, oxidative stability, iodine value, and fat color, remains a challenge that must be addressed. Conventional methods such as visual inspection, mechanical methods, and chemical methods are used off the production line, which often results in an inaccurate representation of the process because the dynamics are lost due to the time required to perform the analysis. Consequently, rapid, and non-destructive alternative methods are needed. In this paper, the traditional fat quality assessment techniques are discussed with emphasis on spectroscopic techniques as an alternative. Potential spectroscopic techniques include infrared spectroscopy, nuclear magnetic resonance and Raman spectroscopy. Hyperspectral imaging as an emerging advanced spectroscopy-based technology is introduced and discussed for the recent development of assessment for fat quality attributes. All techniques are described in terms of their operating principles and the research advances involving their application for pork fat quality parameters. Future trends for the non-destructive spectroscopic techniques are also discussed.

A comparative study of using in-line near-infrared spectra, ultraviolet spectra and fused spectra to monitor Panax notoginseng adsorption process

The step of enriching and purifying saponins by macroporous resin column chromatography is closely related to the safety and efficacy of Panax notoginseng products during their manufacturing processes. Adsorption process is one of the most critical unit operations within each chromatographic cycle. In order to understand the adsorption process directly, it is necessary to develop a rapid and precise method to monitor the adsorption process in real time. In this study, comparative evaluation of using near-infrared (NIR) spectra, ultraviolet (UV) spectra and fused spectra to monitor the adsorption process of P. notoginseng was conducted. The uninformative variable elimination by partial least squares (UVE-PLS) regression models were established for quantification of notoginsenoside R1, ginsenoside Rg1, ginsenoside Re, ginsenoside Rb1 and ginsenoside Rd in effluents based on different spectra. There was a significant improvement provided by the models based on fused spectra. The results in this work were conducive to solving the problems about real-time quantitative analysis of saponins during P. notoginseng adsorption. The fusion method of NIR and UV spectra combined with UVE-PLS regression could be a promising strategy to real-time analyze the components, which are difficult to be quantified by individual spectroscopic technique.

A multi-model fusion strategy for multivariate calibration using near and mid-infrared spectra of samples from brewing industry

Near and mid-infrared (NIR/MIR) spectroscopy techniques have gained great acceptance in the industry due to their multiple applications and versatility. However, a success of application often depends heavily on the construction of accurate and stable calibration models. For this purpose, a simple multi-model fusion strategy is proposed. It is actually the combination of Kohonen self-organizing map (KSOM), mutual information (MI) and partial least squares (PLSs) and therefore named as KMICPLS. It works as follows: First, the original training set is fed into a KSOM for unsupervised clustering of samples, on which a series of training subsets are constructed. Thereafter, on each of the training subsets, a MI spectrum is calculated and only the variables with higher MI values than the mean value are retained, based on which a candidate PLS model is constructed. Finally, a fixed number of PLS models are selected to produce a consensus model. Two NIR/MIR spectral datasets from brewing industry are used for experiments. The results confirms its superior performance to two reference algorithms, i.e., the conventional PLS and genetic algorithm-PLS (GAPLS). It can build more accurate and stable calibration models without increasing the complexity, and can be generalized to other NIR/MIR applications.