Complete Defense Framework to Protect Deep Neural Networks against Adversarial Examples

Table 1

The parameters of evaluated attacks.

Attacks	FGSM	R-FGSM	BIM	UAP	DeepFool	CW_UT	CW_T
				0.1	—	—	—
	——	0.01	1	—	—	—	—
	—	—	—	0.8	—	—	—
Iterations	—	—	20	400	200	300	300