Table 7 Results of FGSM-based adversarial testing on the ADNI dataset

From: Automated Alzheimer’s disease detection using active learning model with reinforcement learning and scope loss function

Model

Accuracy

F-measure

G-measure

AUC

HBOA-MLP19

75.052 ± 0.049

80.621 ± 0.001

81.143 ± 0.066

0.602 ± 0.084

CNN20

76.096 ± 0.073

81.667 ± 0.092

82.151 ± 0.061

0.619 ± 0.046

VGG1621

76.672 ± 0.050

82.907 ± 0.091

83.328 ± 0.079

0.632 ± 0.026

CLADSI22

77.406 ± 0.074

84.315 ± 0.034

84.734 ± 0.020

0.647 ± 0.061

DQN29

78.263 ± 0.008

85.352 ± 0.030

85.737 ± 0.044

0.660 ± 0.075

4D-AlzNet50

79.817 ± 0.095

86.562 ± 0.051

86.971 ± 0.083

0.670 ± 0.033

XAI51

80.377 ± 0.079

87.410 ± 0.041

87.832 ± 0.081

0.682 ± 0.063

RESNET50 KNN52

81.466 ± 0.005

88.725 ± 0.001

89.117 ± 0.065

0.698 ± 0.023

Proposed

86.236 ± 0.056

91.478 ± 0.068

91.852 ± 0.044

0.789 ± 0.059

Back to article page