Table 10 Training time efficiency for pretrained models.

From: A novel technique for ransomware detection using image based dynamic features and transfer learning to address dataset limitations

Model

Without fine-tuning

With fine-tuning (see Table 4)

Accuracy (%)

Avg. training time per sample (s)

Accuracy (%)

Avg. training time per sample (s)

ResNet50

99.91

0.571

99.96

0.440

EfficientNetB0

99.26

0.208

99.78

0.193

InceptionV3

98.77

0.380

97.84

0.098

Xception

99.01

0.549

98.90

0.262

VGG16

98.20

0.932

97.26

0.860

VGG19

97.47

1.460

98.00

1.444

Back to article page