Table 4 Performance Analysis of SRB-VGG19 Dense ANL Model.

From: Restricted Boltzmann machine with Sobel filter dense adversarial noise secured layer framework for flower species recognition

FGSM with different gradient \(\nabla\)

Defended Accuracy with ANL Layer

\(\nabla\)= 0

\(\nabla\)= 2

\(\nabla\)= 4

SRB-VGG19 Dense

97.53

97.11

96.47

SRB-VGG19 Dense ANL with \(\in\) =0.00

98.12

98.14

98.45

SRB-VGG19 Dense ANL with \(\in\) =0.01

98.43

98.34

98.31

SRB-VGG19 Dense ANL with \(\in\) =0.10

98.24

98.35

98.27

SRB-VGG19 Dense ANL with \(\in\) =0.15

98.33

98.24

98.18

SRB-VGG19 Dense ANL with \(\in\) =0.20

98.02

98.12

98.24

SRB-VGG19 Dense ANL with \(\in\) =0.25

98.45

98.36

98.41

Back to article page