Table 7 Accuracy comparison using the NSL-KDD dataset [13]
Paper | Methods | Accuracy (%) | |||
|---|---|---|---|---|---|
DoS | Probe | R2L | U2R | ||
Zhang et al. [32] | Autoencoder (ANN Deep learning) | 83.95 | 80.38 | 11.26 | 32.84 |
Nkiama et al. [19] | ANOVA F-test + RFE | 99.9 | 99.8 | 99.88 | 99.9 |
Revathi and Malathi [33] | CFS + Random Forest | 99.1 | 98.9 | 98.7 | 97.9 |
Benaddi et al. [34] | PCA-Fuzzy Clustering KNN | 94.23 | 78.86 | 80.09 | 69.87 |
Megantara and Ahmad [28] | Feature importance + RFE | 88.98 | 91.18 | 81.29 | 99.42 |
Feature importance + RFE + CV 10 | 99.52 | 98.9 | 94.99 | 99.65 | |
Lian et al. [35] | Ensemble Decision Tree + RFE | 99.74 | 99.2 | 98.21 | 99.77 |
Hussain et al. [31] | Hybrid SVM—ANN | 100 | 99.9 | 77.4 | 88.6 |
Jia et al. [36] | New Deep Neural Network (NDNN) | 98.67 | 97.73 | 96.94 | 81.82 |
Proposed method | Hybrid Machine Learning Method | 99.94 | 99.89 | 99.89 | 99.22 |