Model Extraction Attack Robustness on TinyImageNet
0.3914AccuracyComMark
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ComMarkVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.3914 | 0.9959 | |
| ContentVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.3702 | 0.0322 | |
| MABVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.3697 | 0.13 | |
| NoiseVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.3679 | 0.1485 | |
| AbstractVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.232 | 0.88 | |
| BlindMarkVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.2094 | 0.132 | |
| MEADVictim Model Architecture=ResNet18, Victim Training Dataset=CIFAR10, Adversary Model Architecture=DenseNet161, Adversary Query Set=TinyImageNet2025.12 | 0.1716 | 0.9974 |