Adversarial Image Classification on CIFAR-10 (Blend Attack)
94.89AccuracyBefore
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Beforedefense=None2022.02 | 94.89 | 87.94 | |
| Neural Cleansedefense=Neural Cleanse2022.02 | 94.89 | 87.94 | |
| PBEclean_data_available=true2022.02 | 93.65 | 0.68 | |
| ANPdefense=Adversarial Neuron Pruning2022.02 | 93.12 | 1.24 | |
| PBEclean_data_available=false2022.02 | 92.76 | 0.84 | |
| NADdefense=Neural Attention Distillation2022.02 | 89.91 | 2.38 | |
| Fine-pruningdefense=Fine-pruning2022.02 | 87.96 | 74.15 | |
| Fine-tuningdefense=Fine-tuning2022.02 | 86.6 | 5.36 |