Fig. 11

Confusion matrices for the VGG16 network with various classifiers. The color bars represent TN, FP, FN, and TP in this order. The classifiers are arranged as follows: SVM, RBF, \(\gamma =\)“scale”, C = 15;   SVM, RBF, \(\gamma =10^{-7},\) C = 10; \(\underline{\text {SVM, RBF,} \,\gamma =10^{-7},}\) C = 15; SVM, RBF, \(\gamma =10^{-7},\) C = 20; SVM, polynomial of deg = 2; SVM, polynomial of deg = 3; DT, min−samples−leaf = 2; DT, min−samples−leaf = 3; RF, n−estimators = 100; RF, n−estimators = 200. The extra parameters of the DT are: criterion = “gini”, random−state = 42, and max−depth=3