Image Classification on CIFAR-10 (test) (Adversarial Robustness Metrics)
84.52Clean AccuracyTARO
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| TAROEvaluation type=Self-supervised linear evaluation, SSL Framework=BYOL, Attack Type=Lours-byol2022.10 | 84.52 | — | — | — | — | — | 31.2 | 22.01 | |
| TAROEvaluation type=Self-supervised robust linear evaluation, SSL Framework=SimSiam, Attack Type=Lours-ss2022.10 | 76.19 | — | — | — | — | — | 45.57 | 39.25 | |
| TAROEvaluation type=Self-supervised linear evaluation, SSL Framework=SimSiam, Attack Type=Lours-ss2022.10 | 74.87 | — | — | — | — | — | 44.71 | 36.39 | |
| TAROEvaluation type=Self-supervised robust linear evaluation, SSL Framework=BYOL, Attack Type=Lours-byol2022.10 | 74.33 | — | — | — | — | — | 40.84 | 29.91 | |
| BYOLEvaluation type=Self-supervised linear evaluation, SSL Framework=BYOL, Attack Type=Lbyol2022.10 | 72.65 | — | — | — | — | — | 16.2 | 0.01 | |
| SimSiamEvaluation type=Self-supervised linear evaluation, SSL Framework=SimSiam, Attack Type=Lss2022.10 | 71.78 | — | — | — | — | — | 32.28 | 24.41 | |
| SimSiamEvaluation type=Self-supervised robust linear evaluation, SSL Framework=SimSiam, Attack Type=Lss2022.10 | 68.88 | — | — | — | — | — | 37.84 | 31.44 | |
| BYOLEvaluation type=Self-supervised robust linear evaluation, SSL Framework=BYOL, Attack Type=Lbyol2022.10 | 54.01 | — | — | — | — | — | 27.24 | 4.49 |