Abstract:

Escherichia coli O157:H7, Listeria monocytogenes and Staphylococcus aureus were identified by combined use
of near infrared (NIR) spectroscopy and support vector machine (SVM) in this study. After pretreatment, the NIR spectra
of the three species of food-borne bacteria were analyzed by principal component analysis (PCA), and the first 26 principal
components (PCs) were extracted and used as inputs to establish an SVM model. A comparison was performed between the
radial basis function (RBF) SVM classifiers and the polynomial SVM classifiers. The results indicate that the three kinds
of food-borne bacteria can be completely identified by the RBF-SVM classifier with a nuclear parameter of 0.5, giving a
prediction accuracy of 100% and results consistent with those obtained from the Chinese national standard.

Key words: NIR, food-borne bacteria, SVM, identification