Table 7 Ablation study of the proposed hybrid + SUCMO method.
From: Advances to IoT security using a GRU-CNN deep learning model trained on SUCMO algorithm
Proposed without Feature Extraction | Hybrid + SUCMO | |
|---|---|---|
UNSW-NB15 Dataset | ||
Accuracy | 0.82 | 0.933667 |
Sensitivity | 0.82 | 0.93898 |
Specificity | 0.82 | 0.9486 |
Precision | 0.82 | 0.909379 |
F-measure | 0.82 | 0.936891 |
MCC | 0.64 | 0.868224 |
NPV | 0.82 | 0.841257 |
FPR | 0.18 | 0.001912 |
FNR | 0.18 | 0.100819 |
BoT-IoT Dataset | ||
Accuracy | 0.656033 | 0.929692 |
Sensitivity | 0.140083 | 0.920196 |
Specificity | 0.785021 | 0.929987 |
Precision | 0.140083 | 0.93009 |
F-measure | 0.140083 | 0.927879 |
MCC | 0.074896 | 0.920807 |
NPV | 0.785021 | 0.929797 |
FPR | 0.214979 | 0.041385 |
FNR | 0.859917 | 0.088888 |
