Image Classification on CIFAR-100 linear evaluation
46.54Clean AccuracyRoCL + TARO
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RoCL + TAROEvaluation protocol=linear evaluation, Backbone=ResNet182022.10 | 46.54 | 18.91 | |
| RoCLEvaluation protocol=linear evaluation, Backbone=ResNet182022.10 | 45.99 | 17.17 | |
| SimSiam* + TAROEvaluation protocol=linear evaluation, Backbone=ResNet182022.10 | 36.02 | 22.18 | |
| SimSiam*Evaluation protocol=linear evaluation, Backbone=ResNet18, Note=naïve adversarial training applied in SimSiam2022.10 | 24.43 | 13.34 |