Image Classification on CIFAR-10 (test) (Uniform Noise and Flip Perturbations)
92.92Accuracy (Uniform Noise 0%)EBOMLC
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| EBOMLCBackbone=ResNet-32 (frozen), Training Time=1.00×2026.05 | 92.92 | 92.18 | 89.58 | 83.51 | 76.02 | 90.74 | 90.54 | 1.29 | |
| MLCBackbone=ResNet-32 (frozen), Training Time=2.28×2026.05 | 92.76 | 89.01 | 86.08 | 82.08 | 74.1 | 89.39 | 87.12 | 3.14 | |
| GLCBackbone=ResNet-32, Training Time=0.33×2026.05 | 92.25 | 89.31 | 86.2 | 74.27 | 64.38 | 90.97 | 89.75 | 3 | |
| DivideMixBackbone=ResNet-32, Training Time=1.12×2026.05 | 91.32 | 87.15 | 85.16 | 82.45 | 77.69 | 82.9 | 78.38 | 4.43 | |
| PMW-NetBackbone=ResNet-32, Training Time=1.95×2026.05 | 90.74 | 88.94 | 86.8 | 81.56 | 45.3 | 88.79 | 87.36 | 4.29 | |
| Co-teaching+Backbone=ResNet-32, Training Time=0.51×2026.05 | 89.14 | 85.79 | 82.98 | 56.37 | 44.85 | 77.56 | 72.93 | 7.43 | |
| MW-NetBackbone=ResNet-32, Training Time=1.89×2026.05 | 88.91 | 80.61 | 71.55 | 64.02 | 48.36 | 87.11 | 84.75 | 6.86 | |
| L2RWBackbone=ResNet-32, Training Time=1.69×2026.05 | 85.16 | 82.51 | 78.01 | 72.19 | 55.81 | 84.49 | 80.86 | 6.57 | |
| MLC-DBackbone=ResNet-32 (frozen), k=5, Training Time=2.32×2026.05 | 67.75 | 66.84 | 67.56 | 67.14 | 67.58 | 66.18 | 62.57 | 8 |