Targeted Adversarial Attack on ImageNet (test) - Model Suite VGG/ResNet/Transformer Mix
84.52Model Performance: VGG16CFM+SGM
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| CFM+SGMSource Model=ResNet-50, Number of Iterations=3002024.10 | 84.52 | 83.41 | 89.23 | 89.13 | 78.42 | 32.18 | 58.09 | 66.62 | 71.09 | 71.97 | |
| CFMSource Model=ResNet-50, Number of Iterations=3002024.10 | 81.86 | 81.44 | 87.82 | 88.06 | 76.73 | 23.46 | 50.65 | 61.58 | 69.5 | 69.89 | |
| Logit-Margin+SGMSource Model=ResNet-50, Number of Iterations=3002024.10 | 71.69 | 68.19 | 86.74 | 90.66 | 70.71 | 11.51 | 22.03 | 44.56 | 54.65 | 57.6 | |
| Logit+SGMSource Model=ResNet-50, Number of Iterations=3002024.10 | 63.49 | 59.95 | 77.35 | 84.93 | 65.92 | 10.82 | 20.49 | 41.54 | 52.76 | 55.71 | |
| Logit-MarginSource Model=ResNet-50, Number of Iterations=3002024.10 | 47.86 | 47.07 | 72.44 | 77.58 | 52.77 | 4.01 | 8.85 | 22.31 | 34.72 | 37.08 | |
| LogitSource Model=ResNet-50, Number of Iterations=3002024.10 | 37.9 | 38.08 | 61.39 | 70.03 | 47.28 | 3.41 | 8.03 | 19.74 | 32.12 | 34.26 | |
| FFT+SGMSource Model=ResNet-50, Number of Iterations=3002024.10 | 16.99 | 14.88 | 25.35 | 22.15 | 10.87 | 0.68 | 2.44 | 3.08 | 4.71 | 5.15 | |
| FFTSource Model=ResNet-50, Number of Iterations=3002024.10 | 9.72 | 9.03 | 18.83 | 15.7 | 6.24 | 0.25 | 0.98 | 1.42 | 3.13 | 3.22 |