Research Article
Trust in Intrusion Detection Systems: An Investigation of Performance Analysis for Machine Learning and Deep Learning Models
Table 16
The performance of deep learning for KDD dataset using LSTM and GRU with all features.
| Model | Evaluation metric | Cross-validation performance | Testing performance | DOS | Normal | Probe | R2L | U2R | DOS | Normal | Probe | R2L | U2R |
| LSTM with one layer | TNR | 98.7 | 99.87 | 99.24 | 91.83 | 99.88 | 99.98 | 99.96 | 100.00 | 99.99 | 100.00 | FPR | 1.3 | 0.13 | 0.76 | 8.17 | 0.12 | 0.02 | 0.04 | 0.00 | 0.01 | 0.00 | FNR | 10.42 | 9.64 | 46.16 | 0.68 | 11.67 | 0.01 | 0.06 | 1.24 | 7.62 | 66.67 | Accuracy | 98.46 | 99.79 | 97.47 | 98.63 | 99.68 | 99.99 | 99.95 | 99.99 | 99.97 | 99.99 | Precision | 65.61 | 86.16 | 74.48 | 99.17 | 93.07 | 100.00 | 99.80 | 99.55 | 97.22 | 66.67 | Recall | 89.58 | 90.36 | 53.84 | 99.32 | 88.33 | 99.99 | 99.94 | 98.76 | 92.38 | 33.33 | F-score | 75.6 | 88.2 | 62.28 | 99.24 | 90.62 | 100.00 | 99.87 | 99.15 | 94.74 | 44.44 |
| LSTM with two layer | TNR | 99.89 | 99.75 | 100 | 99.95 | 100 | 99.92 | 99.93 | 99.99 | 99.96 | 100.00 | FPR | 0.11 | 0.25 | 0 | 0.05 | 0 | 0.08 | 0.07 | 0.01 | 0.04 | 0.00 | FNR | 0.18 | 0.35 | 2.66 | 15.57 | 100 | 0.01 | 0.29 | 2.22 | 10.85 | 100.00 | Accuracy | 99.83 | 99.73 | 99.97 | 99.91 | 99.99 | 99.98 | 99.89 | 99.98 | 99.93 | 99.99 | Precision | 99.97 | 98.85 | 99.58 | 82.2 | 0 | 99.98 | 99.70 | 99.37 | 84.92 | 0 | Recall | 99.82 | 99.65 | 97.34 | 84.43 | 0 | 99.99 | 99.71 | 97.78 | 89.15 | 0.00 | F-score | 99.9 | 99.25 | 98.44 | 83.27 | 0 | 99.99 | 99.70 | 98.57 | 86.98 | 0 |
| GRU with one layer | TNR | 99.88 | 99.56 | 100 | 99.96 | 100 | 99.92 | 99.90 | 100.00 | 99.97 | 100.00 | FPR | 0.12 | 0.44 | 0 | 0.04 | 0 | 0.08 | 0.10 | 0.00 | 0.03 | 0.00 | FNR | 0.37 | 0.32 | 3.07 | 14.28 | 100 | 0.04 | 0.25 | 2.93 | 11.73 | 100.00 | Accuracy | 99.67 | 99.58 | 99.97 | 99.92 | 99.99 | 99.95 | 99.87 | 99.97 | 99.93 | 99.99 | Precision | 99.97 | 98 | 99.46 | 84.57 | 0 | 99.98 | 99.53 | 99.54 | 86.74 | 0 | Recall | 99.63 | 99.68 | 96.93 | 85.72 | 0 | 99.96 | 99.75 | 97.07 | 88.27 | 0.00 | F-score | 99.8 | 98.83 | 98.17 | 85.05 | 0 | 99.97 | 99.64 | 98.29 | 87.50 | 0 |
| GRU with two layer | TNR | 99.89 | 99.53 | 99.99 | 99.96 | 100 | 99.96 | 99.87 | 100.00 | 99.97 | 100.00 | FPR | 0.11 | 0.47 | 0.01 | 0.04 | 0 | 0.04 | 0.13 | 0.00 | 0.03 | 0.00 | FNR | 0.38 | 0.28 | 2.62 | 23.67 | 100 | 0.05 | 0.19 | 2.67 | 14.96 | 100.00 | Accuracy | 99.67 | 99.56 | 99.97 | 99.9 | 99.99 | 99.95 | 99.86 | 99.97 | 99.93 | 99.99 | Precision | 99.98 | 97.87 | 99.2 | 86.37 | 0 | 99.99 | 99.43 | 99.64 | 87.35 | 0 | Recall | 99.62 | 99.72 | 97.38 | 76.33 | 0 | 99.95 | 99.81 | 97.33 | 85.04 | 0.00 | F-score | 99.8 | 98.78 | 98.28 | 76.73 | 0 | 99.97 | 99.62 | 98.47 | 86.18 | 0 |
|
|