Image Classification on CIFAR-100 symmetric noise
77.3Accuracy (20% Noise, Best)DivideMix
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DivideMix2020.02 | 77.3 | 76.9 | 74.6 | 74.2 | 60.2 | 59.6 | 31.5 | 31 | |
| M-correction2020.02 | 73.9 | 73.4 | 66.1 | 65.4 | 48.2 | 47.6 | 24.3 | 20.5 | |
| P-correctionre-implementation=true2020.02 | 69.4 | 68.1 | 57.5 | 56.4 | 31.1 | 20.7 | 15.3 | 8.8 | |
| Meta-Learningre-implementation=true2020.02 | 68.5 | 67.7 | 59.2 | 58 | 42.4 | 40.1 | 19.5 | 14.3 | |
| Mixup2020.02 | 67.8 | 66 | 57.3 | 46.6 | 30.8 | 17.6 | 14.6 | 8.1 | |
| Co-teaching+re-implementation=true2020.02 | 65.6 | 64.1 | 51.8 | 45.3 | 27.9 | 15.5 | 13.7 | 8.8 | |
| Bootstrap2020.02 | 62.1 | 62 | 46.6 | 37.9 | 19.9 | 8.9 | 10.2 | 3.8 | |
| Cross-Entropy2020.02 | 62 | 61.8 | 46.7 | 32.3 | 19.9 | 8.8 | 10.1 | 3.5 | |
| F-correction2020.02 | 61.5 | 61.4 | 46.6 | 37.3 | 19.9 | 9 | 10.2 | 3.4 |