Adversarial Detection on ImageNet MIFGSM attack (test)
99.31AUROCPerturbation Forgery
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Perturbation ForgeryDetect Attacks=Gradient + GAN + Diffusion, Model-Agnostic=true2024.05 | 99.31 | 4.85 | |
| EPSADDetect Attacks=Gradient, Model-Agnostic=true2024.05 | 99.18 | 396.81 | |
| SPADDetect Attacks=Gradient + GAN, Model-Agnostic=true2024.05 | 98.2 | 4.56 | |
| LIDDetect Attacks=Gradient, Model-Agnostic=false2024.05 | 91.46 | 1.8 | |
| LiBReDetect Attacks=Gradient, Model-Agnostic=false2024.05 | 87.25 | 2.56 |