Table 10 Comparison of metrics for different feature extraction modules.
From: Multimodal fusion based few-shot network intrusion detection system
Feature extraction module | ACC (%) | FLOPs (M) | Params (M) | ||
|---|---|---|---|---|---|
K = 5 | K = 10 | K = 15 | |||
CNN2D | 83.6 | 89.3 | 91.4 | 38.97 | 0.13 |
CNN3D | 84.4 | 92.3 | 89.7 | 43.69 | 11.24 |
ResNet18 | 88.4 | 90.0 | 92.6 | 142.43 | 11.17 |
VGG16 | 20.0 | 20.0 | 20.0 | 1367.65 | 134.27 |
AlexNet | 82.2 | 92.5 | 93.3 | 90.59 | 57.01 |
GoogleNet | 71.1 | 94.5 | 93.7 | 116.61 | 5.60 |
DenseNet | 77.3 | 83.5 | 91.4 | 229.98 | 6.95 |
MobileNet | 71.4 | 83.4 | 92.0 | 26.05 | 2.23 |
SqueezeNet | 64.0 | 70.2 | 75.3 | 22.29 | 0.72 |
Self-Sufficient | 92.8 | 92.9 | 93.1 | 43.99 | 11.45 |
Transfer-Enhanced | 93.4 | 95.2 | 98.8 | 43.99 | 11.45 |
