Model Extraction Attack Robustness on CIFAR100
19.49AccuracyMEAD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MEADVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 19.49 | 98.43 | |
| BlindMarkVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 18.67 | 11.8 | |
| ComMarkVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 17.74 | 97.61 | |
| ContentVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 14.44 | 6.6 | |
| AbstractVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 10 | 10 | |
| NoiseVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 10 | 94.96 | |
| MABVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=VGG16, Adversary Query Set=CIFAR1002025.12 | 10 | 11 |