Twitter Spam Detection (test)
90.11Certification S.R.NU-δ-MDt
Evaluation Results
| Method | Links | |
|---|---|---|
| NU-δ-MDtCertification approach=NUC-MDt, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 90.11 | |
| NU-δ-MDtCertification approach=NUC-MD, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 85.34 | |
| NU-δ-MDtCertification approach=NUC-SHAP, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 71.25 | |
| NU-δ-MDtCertification approach=NUC-Pearson, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 67.14 | |
| Uniform-δCertification approach=NUC-MDt, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 66.45 | |
| Uniform-δCertification approach=NUC-MD, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 65.72 | |
| Uniform-δCertification approach=NUC-SHAP, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 64.24 | |
| NU-δ-MDtCertification approach=UC (Standard Randomized Smoothing), Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 61.8 | |
| Uniform-δCertification approach=NUC-Pearson, Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 61.11 | |
| Uniform-δCertification approach=UC (Standard Randomized Smoothing), Training distortion budget (||δ||2)=5, Certification distortion budget (||δ||2)=2.82021.02 | 50.96 |