Table 2.

Effects of the setting parameters for training the CNNs on their performance (the values followed by ± indicate the standard deviation)

No of training epochs	30	40	50	30	40	50
AlexNet				VGG19
	VA (%)			VA (%)
No of mini-baches
5	88.89 ± 4.18	92.22 ± 4.84	93.15 ± 2.48	91.85 ± 2.79	90.33 ± 4.22	92.78 ± 3.54
10	90.00 ± 3.72	91.30 ± 3.15	92.78 ± 4.14	91.48 ± 4.26	92.18 ± 3.44	90.94 ± 4.76
20	93.33 ± 1.99	92.22 ± 2.44	92.59 ± 2.89	90.81 ± 2.97	90.13 ± 5.48	89.63 ± 3.51
	Loss			Loss
5	0.31 ± 0.14	0.19 ± 0.07	0.20 ± 0.06	0.23 ± 0.15	0.27 ± 0.15	0.27 ± 0.23
10	0.22 ± 0.09	0.21 ± 0.11	0.19 ± 0.11	0.27 ± 0.21	0.14 ± 0.08	0.18 ± 0.11
20	0.10 ± 0.03	0.09 ± 0.04	0.12 ± 0.05	0.19 ± 0.14	0.21 ± 0.16	0.19 ± 0.13
	Training time (s/epoch)			Training time (s/epoch)
5	11.82 ± 0.49	10.92 ± 0.25	10.74 ± 0.19	148.08 ± 14.39	150.4 ± 4.34	140.22 ± 8.45
10	12.84 ± 2.55	15.78 ± 0.40	15.90 ± 0.51	197.76 ± 1.38	192.40 ± 21.64	188.88 ± 4.45
20	25.02 ± 0.29	25.62 ± 1.33	27.48 ± 1.26	263.33 ± 21.15	282.34 ± 27.18	276.54 ± 35.90

VA validation accuracy, Loss the error of the model