In the classification (identification) phase as shown in Fig. 5,
total of the adulterated sesame oil samples are randomly split into
two sets, one used as calibration set including 580 samples and the
other with 160 adulterated sesame oils used as validation set (randomly
selected from each class according to the ratio 8/37). Here,
the discrimination of five kinds of mixed oils are also tried to be
done using SVM model with RBF kernel. Through the two-step
parameter optimization, the optimal parameters are determined
with the values of c = 137.2 and g = 0.0583, and the number of support
vectors is 75 (only 12.7% of total samples in calibration set).
With this SVM classification model, the correct predictions are
obtained for the 160 samples, i.e. 100%. These results show that
the SVM classification can further identify which kind of vegetable
oil is adulterated with the sesame oil.
In the classification (identification) phase as shown in Fig. 5,total of the adulterated sesame oil samples are randomly split intotwo sets, one used as calibration set including 580 samples and theother with 160 adulterated sesame oils used as validation set (randomlyselected from each class according to the ratio 8/37). Here,the discrimination of five kinds of mixed oils are also tried to bedone using SVM model with RBF kernel. Through the two-stepparameter optimization, the optimal parameters are determinedwith the values of c = 137.2 and g = 0.0583, and the number of supportvectors is 75 (only 12.7% of total samples in calibration set).With this SVM classification model, the correct predictions areobtained for the 160 samples, i.e. 100%. These results show thatthe SVM classification can further identify which kind of vegetableoil is adulterated with the sesame oil.
การแปล กรุณารอสักครู่..
