Improving the Performance of Deep Learning Model-Based Classification by the Analysis of Local Probability

Table 6

The result of random probability on CIFAR-100 and Mini-ImageNet with fusion operators.

Local probability	Methods
	ResNeSt50 [26]	VGG16 (%) [25]	VoVNet-57 (%) [24]	CC-KL trainable (%) [27]	CC-KL trainable with our framework (%)
CIFAR-100, Rand (0, 1)	41.93	46.19	63.98	65.52	66.27
CIFAR-100, Rand (−1, 1)	41.83	46.21	63.78	65.57	74.93
CIFAR-100, Rand (−2, 1)	42.03	46.35	63.42	66.16	77.02
Mini-ImageNet, Rand (0, 1)	40.34	44.54	72.65	73.39	74.55
Mini-ImageNet, Rand (−1, 1)	40.64	44.84	72.58	73.47	81.23
Mini-ImageNet, Rand (−2, 1)	41.04	45.01	72.47	74.45	85.48