ColoRectalCADx: Expeditious Recognition of Colorectal Cancer with Integrated Convolutional Neural Networks and Visual Explanations Using Mixed Dataset Evidence

Table 8

Comparison of accuracies of CNN+SVM architectures.

Individual CNN	CNN+SVM training accuracy	CNN+SVM testing accuracy
AlexNet	44.33	45
DarkNet-19	77.69	72
ResNet-50v2	78.27	78
DenseNet-201	74.29	73
EfficientNet-B7	48.15	60
VGG-16	70.07	65
VGG-19	63.96	65
NasNetLarge	73.47	68
InceptionResNetV2	78.1	69
Integrated CNN	CNN+SVM training accuracy	CNN+SVM testing accuracy
ADaRDEV^2I-22	67.92	66
ADaRDEV^2-22	69.45	57
RDEV^2-22	69.51	69
ADaDR-22	82.04	75
ADaR-22	63.31	64
DaRD-22	68.68	66
DEV-22	61.82	67
ADa-22	58.21	51
RD-22	61.81	64
RV-22	30.31	30
AD-22	66.47	65
DaR-22	60.53	59