Research Article
Dark Web Data Classification Using Neural Network
Table 3
Performance of models (SVM, TSVM, NN-S3VM).
| Database | SVM | TSVM | NN- S3VM | Accuracy (%) | MCC | ROC area | Accuracy (%) | MCC | ROC area | Accuracy (%) | MCC | ROC area |
| CIRA-CIC-DoHBrw-2020 | 0.61 | 0.54 | 0.72 | 0.65 | 0.34 | 0.74 | 0.81 | 0.55 | 0.82 | CSE-CIC-IDS2018 on AWS | 0.63 | 0.28 | 0.82 | 0.64 | 0.31 | 0.83 | 0.84 | 0.34 | 0.83 | Intrusion detection evaluation dataset (CIC-IDS2017) | 0.64 | 0.30 | 0.92 | 0.64 | 0.25 | 0.93 | 0.84 | 0.29 | 0.94 | Intrusion detection evaluation dataset (ISCXIDS2012) | 0.71 | 0.50 | 0.91 | 0.76 | 0.49 | 0.92 | 0.87 | 0.13 | 0.95 | DDoS evaluation dataset (CIC-DDoS2019) | 0.74 | 0.30 | 0.92 | 0.78 | 0.29 | 0.94 | 0.91 | 0.13 | 0.96 | Investigation of the android malware (CIC-InvesAndMal2019) | 0.73 | 0.31 | 0.93 | 0.79 | 0.34 | 0.95 | 0.93 | 0.18 | 0.91 | Android botnet dataset | 0.71 | 0.51 | 0.94 | 0.81 | 0.41 | 0.95 | 0.94 | 0.46 | 0.97 |
|
|