Research Article
Network Traffic Anomaly Detection Model Based on Feature Reduction and Bidirectional LSTM Neural Network Optimization
Table 9
Detection results of the four models.
| Dataset | Model | Evaluation metrics | Accuracy (%) | Precision (%) | Recall (%) | F-score (%) |
| NSL-KDD | FR-APPSO-LSTM | 90.02 | 83.33 | 95.26 | 88.89 | FR-BiLSTM | 90.81 | 84.46 | 97.54 | 90.53 | APPSO-BiLSTM | 90.40 | 83.86 | 95.97 | 89.61 | FR-APPSO-BiLSTM | 91.76 | 85.37 | 98.50 | 91.46 |
| UNSW-NB15 | FR-APPSO-LSTM | 85.19 | 92.97 | 92.26 | 92.61 | FR-BiLSTM | 89.84 | 97.01 | 97.23 | 97.12 | APPSO-BiLSTM | 91.41 | 96.92 | 97.00 | 97.06 | FR-APPSO-BiLSTM | 92.08 | 97.88 | 98.32 | 98.10 |
| CICIDS-2017 | FR-APPSO-LSTM | 92.29 | 86.15 | 87.58 | 86.86 | FR-BiLSTM | 93.95 | 98.25 | 97.79 | 98.02 | APPSO-BiLSTM | 94.20 | 98.02 | 97.73 | 98.12 | FR-APPSO-BiLSTM | 95.44 | 98.58 | 98.40 | 98.49 |
|
|