Image Classification Algorithm Based on Deep Learning-Kernel Function

Table 1

Different methods identify accuracy at various training set sizes (unit:%).

Method type	The proportion of training sets.
	20%	30%	40%	70%
HUSVM [47]	82.2	86.8	91.6	93.8
ScSPM [48]	88.1	90.1	92.4	94.5
AlexNet [49]	90.5	92.3	94.5	97.2
VGG + FCNet [50]	91.4	94.7	95.2	98.1
Ours	94.5	96.1	97.8	99.2