Terahertz Spectroscopy Combined with Machine Learning to Identify the Origin of Ginseng

Abstract

An effective identification method of ginseng origin based on terahertz spectroscopy and machine learning was proposed. Extinction coefficients at terahertz frequency of ginseng collected from China, Canada and America are investigated using the proposed method, in the frequency range from 0.2 to 1.3 THz. Principal component analysis is employed to reduce the dimensionality of the original data and extract features. The processed data by principal component analysis is fed into a classification model established by random forest and support vector machines. The identification rate of two classification models can be achieved 88.5% and 87.5% respectively. Compared with SVM model, RF model can obtain better results. Ginseng from America and Canada is more similar and there are relatively high mistaken between Canada and America with both models. The results demonstrating that terahertz spectroscopy combined with machine learning can be used for the identification of the origin of ginseng.