Image Classification on CIFAR-10 (test) (Uniform and Targeted Corruption Evaluation)
87.78Accuracy (Clean)Random
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| RandomBackbone=VGG-19, Keep fraction (alpha)=0.252026.05 | 87.78 | 81.84 | 82.81 | 74.66 | |
| ForgettingBackbone=VGG-19, Keep fraction (alpha)=0.252026.05 | 87.39 | 82.2 | 81.99 | 72.35 | |
| DR-ISBackbone=VGG-19, Keep fraction (alpha)=0.252026.05 | 87.14 | 85.98 | 83.43 | 81.86 | |
| EL2N (ensemble)Backbone=VGG-19, Keep fraction (alpha)=0.252026.05 | 83.94 | 27.53 | 48.62 | 9.39 | |
| Consensus-lossBackbone=VGG-19, Keep fraction (alpha)=0.252026.05 | 81 | 28.2 | 44.29 | 9 | |
| AUM (top-α)Backbone=VGG-19, Keep fraction (alpha)=0.252026.05 | 80.6 | 82.86 | 76.75 | 79.69 |