Image Classification on CIFAR-10 (test) (Threshold Accuracy)
90.56Accuracy (p=0.2)MW-net
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MW-netNoise type=Flip32022.11 | 90.56 | 88.49 | 85.65 | 22.69 | |
| MFRWNoise type=Flip32022.11 | 90.31 | 88.96 | 87.73 | 75.53 | |
| CrossEntropyNoise type=Flip3, Implementation=Author's implementation2022.11 | 90.13 | 88.44 | 82.31 | 20.34 |