Research Article
An Intrusion Detection Method Based on Fully Connected Recurrent Neural Network
Table 4
Accuracy of models with different structures and Learningrates (multiclassification).
| | | KDDTrain+ | KDDTest+ | KDDTest−21(%) |
| HiddenNodes = 60,
Learningrate = 0.1 | 99.84% | 79.87% | 61.98 | HiddenNodes = 60,
Learningrate = 0.5 | 99.87% | 77.46% | 57.18 | HiddenNodes = 60,
Learningrate = 0.8 | 91.23% | 69.29% | 41.85 | HiddenNodes = 80,
Learningrate = 0.1 | 99.82% | 77.73% | 57.84 | HiddenNodes = 80,
Learningrate = 0.5 | 99.53% | 81.29% | 64.67 | HiddenNodes = 80,
Learningrate = 0.8 | 98.97% | 77.09% | 56.64 | HiddenNodes = 120,
Learningrate = 0.1 | 99.85% | 77.02% | 56.55 | HiddenNodes = 120,
Learningrate = 0.5 | 99.87% | 79.44% | 61.11 | HiddenNodes = 120,
Learningrate = 0.8 | 93.90% | 70.32% | 45.49 | HiddenNodes = 160,
Learningrate = 0.1 | 99.68% | 77.85% | 58.05 | HiddenNodes = 160,
Learningrate = 0.5 | 99.80% | 78.73% | 59.71 | HiddenNodes = 160,
Learningrate = 0.8 | 92.79% | 71.15% | 45.64 |
|
|
