Journal of Big Data

Table 3 The average accuracy of the last epoch for each network with different optimizers after 20 epoch (%)

From: Classification of mastoid air cells by CT scan images using deep learning method

Name of architectures		Optimizers
Name of architectures		SGD	RMSprop	AdaGrad	AdaDelta	Adam	AdaMax	Nadam
1	Xception	83.24	82.16	82.31	85.30	82.68	85.13	81.29
2	VGG16	81.47	72.26	70.48	69.27	48.15	78.39	48.15
3	VGG19	81.73	68.47	69.92	60.73	48.15	74.03	48.15
4	ResNet50	80.42	74.03	73.63	77.34	77.50	79.83	73.87
5	ResNet101	81.35	68.31	75.24	77.82	76.77	77.02	77.18
6	ResNet152	80.65	73.39	76.13	76.13	75.81	76.69	75.40
7	ResNet50V2	81.24	73.79	77.02	76.37	75.32	78.16	74.44
8	ResNet101V2	80.45	68.87	76.81	77.21	76.61	78.34	75.73
9	ResNet152V2	80.36	70.73	67.10	70.65	70.48	77.74	70.81
10	InceptionV3	81.40	78.84	80.18	82.53	80.11	85.58	77.02
11	InceptionResNetV2	82.45	81.26	82.14	84.37	82.05	82.42	79.93
12	MobileNet	75.48	79.03	80.55	81.34	78.87	82.24	80.19
13	MobileNetV2	76.85	78.31	76.05	80.29	78.52	79.61	77.11
14	DenseNet121	82.48	76.29	75.00	80.98	81.08	81.70	73.95
15	DenseNet169	82.42	76.05	73.39	80.07	77.42	81.68	75.08
16	DenseNet201	84.22	76.69	77.50	81.54	77.50	80.48	77.02

Back to article page