We report on the analysis of volatile compounds by SPME-GC-MS for individual roasted coffee beans. The aim was to understand the relative abundance and variability of volatile compounds between individual roasted coffee beans at constant roasting conditions. Twenty-five batches of Arabica and robusta species were sampled from 13 countries, and 10 single coffee beans randomly selected from each batch were individually roasted in a fluidised-bed roaster at 210 °C for 3 min. High variability (CV = 14.0–53.3%) of 50 volatile compounds in roasted coffee was obtained within batches (10 beans per batch). Phenols and heterocyclic nitrogen compounds generally had higher intra-batch variation, while ketones were the most uniform compounds (CV < 20%). The variation between batches was much higher, with the CV ranging from 15.6 to 179.3%. The highest variation was observed for 2,3-butanediol, 3-ethylpyridine and hexanal. It was also possible to build classification models based on geographical origin, obtaining 99.5% and 90.8% accuracy using LDA or MLR classifiers respectively, and classification between Arabica and robusta beans. These results give further insight into natural variation of coffee aroma and could be used to obtain higher quality and more consistent final products. Our results suggest that coffee volatile concentration is also influenced by other factors than simply the roasting degree, especially green coffee composition, which is in turn influenced by the coffee species, geographical origin, ripening stage and pre- and post-harvest processing.
In this study, we investigate the use of convolutional neural networks (CNN) for near infrared (NIR) calibration. We propose a unified CNN structure that can be used for general multivariate regression purpose. The comparison between the CNN method and the partial least squares regression (PLSR) method was done on three different NIR datasets of spectra and lab reference values. Datasets are from different sources and contain 6998, 1000 and 415 training and 618, 597 and 108 validation samples, respectively. Results indicated that compared to the PLSR models, the CNN models are more accurate and less noisy. The convolutional layer in the CNN model can automatically find the suitable spectral preprocessing filter on the dataset, which significantly saves efforts in training the model.
This paper investigates the use of least squares support vector machines(LS-SVMs) and Gaussian process regression(GPR) for multivariate spectroscopic calibration. The performances of these two non-linear regression models are assessed and compared to the traditional linear regression model, partial least squares regression(PLSR) on an agricultural example. The non-linear models, LS-SVMs and GPR, showed enhanced generalization ability, especially in maintaining homogeneous prediction accuracy over the range. The two non-linear models generally have similar prediction performance, but showed different features in some situations, especially when the size of the training set varies. This is due to fundamental differences in fitting criteria between these models.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.