Table 6 Class-wise quantitative results specific to traffic types on UNSW-NB15.
From: A novel multi-scale network intrusion detection model with transformer
| Â | Accuracy | Recall | ||||||
|---|---|---|---|---|---|---|---|---|
Potluri et al. 56 | Hooshmand et al. 55 | DNN 58 | Ours | DRaNN 57 | Ashiku et al. 59 | DNN 58 | Ours | |
Analysis | 0.0 | 99.0 | 99.5 | 98.7 | 98.2 | 89.5 | 0.0 | 98.5 |
Backdoor | 0.0 | 12.0 | 95.1 | 99.1 | 98.8 | 91.2 | 34.4 | 98.1 |
Dos | 0.0 | 10.5 | 99.4 | 98.4 | 98.8 | 94.6 | 97.7 | 99.0 |
Exploits | 61.8 | 30.0 | 89.9 | 97.2 | 98.8 | 94.2 | 1.3 | 99.4 |
Fuzzers | 6.8 | 69.5 | 99.9 | 97.4 | 97.1 | 88.6 | 0.0 | 99.0 |
Generic | 97.7 | 69.1 | 78.3 | 98.6 | 99.8 | 95.1 | 57.1 | 99.2 |
Normal | 99.7 | 99.0 | 78.9 | 99.9 | - | 97.2 | 92.8 | 99.7 |
Reconnaisance | 0.0 | 77.2 | 92.7 | 97.0 | 99.2 | 95.1 | 1.8 | 99.7 |
Shell code | 0.0 | 85.0 | 99.0 | 97.5 | 97.8 | 91.6 | 0.0 | 99.0 |
Worms | 0.0 | 76.9 | 98.8 | 98.6 | 98.1 | 89.8 | 0.0 | 99.4 |
