Table 5.

Validation true positive rate (TPR) of CNN models using 3 classes (E: Extrasystole; M: Murmur; N: Normal) and 3 folds without selection process.

Model	TPR
	Fold1				Fold2				Fold3				Avg
	E	M	N	Avg	E	M	N	Avg	E	M	N	Avg
VGG16	0.36	0.77	0.90	0.68	0.62	0.80	0.89	0.77	0.31	0.83	0.93	0.69	0.72
VGG19	0.44	0.80	0.88	0.70	0.54	0.81	0.89	0.75	0.4	0.88	0.91	0.73	0.73
Xception	0.0	0.52	0.77	0.43	0.0	0.41	0.91	0.44	0.0	0.46	0.88	0.44	0.44
ResNet152V2	0.25	0.81	0.65	0.57	0.11	0.69	0.86	0.55	0.28	0.7	0.78	0.59	0.57
ResNet152	0.27	0.70	0.89	0.62	0.25	0.69	0.91	0.62	0.14	0.77	0.83	0.58	0.61
ResNet101V2	0.02	0.63	0.88	0.51	0.14	0.64	0.86	0.55	0.25	0.73	0.79	0.59	0.55
ResNet101	0.16	0.81	0.74	0.57	0.25	0.79	0.80	0.61	0.0	0.79	0.91	0.56	0.58
ResNet50v2	0.33	0.77	0.71	0.60	0.17	0.75	0.79	0.57	0.05	0.56	0.89	0.50	0.56
ResNet50	0.19	0.74	0.85	0.59	0.2	0.73	0.89	0.60	0.17	0.7	0.90	0.59	0.59
NasNetMobile	0.22	0.58	0.81	0.54	0.0	0.6	0.82	0.47	0.0	0.56	0.83	0.46	0.49
MobileNetV2	0.16	0.66	0.84	0.56	0.11	0.67	0.83	0.53	0.14	0.7	0.72	0.52	0.54
MobileNet	0.22	0.65	0.80	0.56	0.22	0.59	0.87	0.56	0.08	0.74	0.78	0.53	0.55
Inceptionv3	0.0	0.65	0.91	0.52	0.0	0.63	0.94	0.52	0.02	0.6	0.95	0.52	0.52
InceptionResNetV2	0.0	0.44	0.91	0.45	0.0	0.69	0.84	0.51	0.0	0.6	0.81	0.47	0.48
DenseNet201	0.25	0.69	0.88	0.60	0.34	0.73	0.86	0.64	0.17	0.73	0.81	0.57	0.60
DenseNet169	0.25	0.67	0.82	0.58	0.11	0.82	0.77	0.57	0.08	0.70	0.88	0.55	0.57
DenseNet121	0.19	0.72	0.83	0.58	0.4	0.72	0.84	0.65	0.17	0.79	0.78	0.58	0.60