Image Classification on CIFAR-10 (test) (AA/SA Metrics)
50.6AADynACL-AIR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DynACL-AIRPre-training=DynACL-AIR, Finetuning=Adversarial Full Finetuning (AFF), Backbone=ResNet-182023.04 | 50.6 | 82.14 | |
| DynACLPre-training=DynACL, Finetuning=Adversarial Full Finetuning (AFF), Backbone=ResNet-182023.04 | 50.52 | 81.86 | |
| ACL-AIRPre-training=ACL-AIR, Finetuning=Adversarial Full Finetuning (AFF), Backbone=ResNet-182023.04 | 49.84 | 82.42 | |
| ACLPre-training=ACL, Finetuning=Adversarial Full Finetuning (AFF), Backbone=ResNet-182023.04 | 49.42 | 82.14 | |
| DynACL-AIRPre-training=DynACL-AIR, Finetuning=Adversarial Linear Finetuning (ALF), Backbone=ResNet-182023.04 | 46.01 | 77.42 | |
| DynACLPre-training=DynACL, Finetuning=Adversarial Linear Finetuning (ALF), Backbone=ResNet-182023.04 | 45.65 | 72.9 | |
| DynACL-AIRPre-training=DynACL-AIR, Finetuning=Standard Linear Finetuning (SLF), Backbone=ResNet-182023.04 | 45.17 | 78.08 | |
| DynACLPre-training=DynACL, Finetuning=Standard Linear Finetuning (SLF), Backbone=ResNet-182023.04 | 45.05 | 75.39 | |
| ACL-AIRPre-training=ACL-AIR, Finetuning=Adversarial Linear Finetuning (ALF), Backbone=ResNet-182023.04 | 41.39 | 78.29 | |
| ACLPre-training=ACL, Finetuning=Adversarial Linear Finetuning (ALF), Backbone=ResNet-182023.04 | 40.61 | 75.56 | |
| ACL-AIRPre-training=ACL-AIR, Finetuning=Standard Linear Finetuning (SLF), Backbone=ResNet-182023.04 | 38.89 | 80.03 | |
| ACLPre-training=ACL, Finetuning=Standard Linear Finetuning (SLF), Backbone=ResNet-182023.04 | 37.39 | 78.27 |