Image Classification on MNIST synthetic label noise (test)
98.74Accuracy (Symmetric, 20%)VolMinNet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| VolMinNet2021.02 | 98.74 | 98.23 | 99.01 | 99 | |
| T-Revision2021.02 | 98.72 | 98.23 | 98.89 | 84.56 | |
| DMI2021.02 | 98.7 | 98.12 | 98.84 | 97.92 | |
| Forward2021.02 | 98.6 | 97.77 | 98.84 | 75.06 | |
| Dual T2021.02 | 98.43 | 98.15 | 98.86 | 96.71 | |
| MentorNet2021.02 | 97.21 | 95.56 | 96.89 | 91.98 | |
| Co-teaching2021.02 | 97.07 | 95.2 | 97 | 96.25 | |
| Decoupling2021.02 | 97.04 | 94.58 | 96.93 | 94.34 |