Table 16. Comparison of recent intrusion detection approaches on the CSE-CIC-IDS2018 dataset (training time on whole dataset and testing time on single data sample).
| Study | Method | Performance measures (%) | No of features | Time required (s) | Model size | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| Feature selection | Classifier | Accuracy | Precision | Recall | F1-score | Training time | Testing time | |||
| Proposed | RF-RFE | ML ensemble | 99.9 | 99.9 | 99.89 | 99.89 | 15 | ~40.76 | 0.007 | ~912 kb |
| CNN | 99.2 | 99.2 | 99.1 | 99.1 | ~1,337 | 0.091 | ~220 kb | |||
| RNN | 99.9 | 99.7 | 99.8 | 99.7 | ~1,450 | 0.095 | ~240 kb | |||
| LSTM | 99.9 | 99.8 | 99.7 | 99.8 | ~2,050 | 0.10 | ~444 kb | |||
| Abdel-Basset et al. (2021) | Traffic Attention |
ResNet | 98.71 | 94.91 | 94.3 | 94.92 | 20 | ~213 | 1.2 | – |
| de Souza, Westphall & Machado (2022) | Extra Tree | RF + DNN + ET | 98.21 | 97.90 | – | – | – | – | 7.19 | – |
| Kim et al. (2020) | CNN | Fully connected network | 91.5 | 70.87 | 83.5 | 76.66 | – | – | – | – |