Anomaly Detection on CICIDS2017 (Security Metrics)
85.99After-Attack AccuracyNetDiffuser
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| NetDiffuserModel Architecture=CNN-2L, Attack Variant=ACG, Before-Attack Accuracy=95.53%2026.03 | 85.99 | 67.24 | 29.61 | 0.7082 | |
| NetDiffuserModel Architecture=CNN-3L, Attack Variant=ACG, Before-Attack Accuracy=96.52%2026.03 | 85.35 | 72.54 | 24.84 | 0.7553 | |
| NetDiffuserModel Architecture=CNN-3L, Attack Variant=PGD, Before-Attack Accuracy=96.52%2026.03 | 84.1 | 70.91 | 25.42 | 0.7516 | |
| NetDiffuserModel Architecture=CNN-3L, Attack Variant=FGSM, Before-Attack Accuracy=96.52%2026.03 | 79.3 | 71.69 | 25.72 | 0.7809 | |
| NetDiffuserModel Architecture=CNN-2L, Attack Variant=PGD, Before-Attack Accuracy=95.53%2026.03 | 78.19 | 60.43 | 36.74 | 0.7303 | |
| NetDiffuserModel Architecture=CNN-2L, Attack Variant=FGSM, Before-Attack Accuracy=95.53%2026.03 | 73.84 | 63.5 | 33.53 | 0.7648 | |
| NetDiffuserModel Architecture=MLP-1L, Attack Variant=PGD, Before-Attack Accuracy=95.30%2026.03 | 71.98 | 69.85 | 26.71 | 0.8556 | |
| NetDiffuserModel Architecture=MLP-5L, Attack Variant=ACG, Before-Attack Accuracy=96.51%2026.03 | 70.59 | 78.65 | 18.51 | 0.8669 | |
| NetDiffuserModel Architecture=MLP-1L, Attack Variant=FGSM, Before-Attack Accuracy=95.30%2026.03 | 70.1 | 70.29 | 26.24 | 0.8579 | |
| NetDiffuserModel Architecture=MLP-1L, Attack Variant=ACG, Before-Attack Accuracy=95.30%2026.03 | 68.76 | 72.72 | 23.7 | 0.8866 | |
| NetDiffuserModel Architecture=MLP-5L, Attack Variant=FGSM, Before-Attack Accuracy=96.51%2026.03 | 63.19 | 71.66 | 25.74 | 0.8616 | |
| NetDiffuserModel Architecture=MLP-5L, Attack Variant=PGD, Before-Attack Accuracy=96.51%2026.03 | 59.38 | 74.81 | 22.48 | 0.8898 | |
| ACGModel Architecture=MLP-1L, Before-Attack Accuracy=95.30%2026.03 | 49.58 | 78.93 | 17.18 | 0.9393 | |
| ACGModel Architecture=MLP-5L, Before-Attack Accuracy=96.51%2026.03 | 49.07 | 78.77 | 18.38 | 0.9291 | |
| FGSMModel Architecture=CNN-3L, Before-Attack Accuracy=96.52%2026.03 | 28.09 | 72.52 | 24.87 | 0.8433 | |
| FGSMModel Architecture=MLP-5L, Before-Attack Accuracy=96.51%2026.03 | 13.35 | 75.68 | 21.59 | 0.9132 | |
| FGSMModel Architecture=MLP-1L, Before-Attack Accuracy=95.30%2026.03 | 5.59 | 70.96 | 25.54 | 0.8898 | |
| PGDModel Architecture=MLP-1L, Before-Attack Accuracy=95.30%2026.03 | 5.59 | 70.14 | 26.4 | 0.8991 | |
| ACGModel Architecture=CNN-2L, Before-Attack Accuracy=95.53%2026.03 | 5.57 | 75.9 | 20.55 | 0.9232 | |
| PGDModel Architecture=MLP-5L, Before-Attack Accuracy=96.51%2026.03 | 5.39 | 76.51 | 20.72 | 0.9105 | |
| FGSMModel Architecture=CNN-2L, Before-Attack Accuracy=95.53%2026.03 | 5.32 | 73.98 | 22.56 | 0.9127 | |
| ACGModel Architecture=CNN-3L, Before-Attack Accuracy=96.52%2026.03 | 5.13 | 75.39 | 21.9 | 0.9124 | |
| PGDModel Architecture=CNN-2L, Before-Attack Accuracy=95.53%2026.03 | 4.47 | 75.55 | 20.91 | 0.9048 | |
| PGDModel Architecture=CNN-3L, Before-Attack Accuracy=96.52%2026.03 | 3.49 | 79.55 | 17.59 | 0.9236 |