Image Classification on CIFAR-10 clean (test) (Epsilon-Dependent Accuracy)
95.42Accuracy (ε=0.2)CORES2*
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CORES2*Noise type=Instance-based, Variant=Star2020.10 | 95.42 | 88.45 | 85.53 | |
| CORES2Noise type=Instance-based2020.10 | 89.5 | 82.84 | 79.66 | |
| Peer LossNoise type=Instance-based2020.10 | 89.33 | 81.09 | 73.73 | |
| SCENoise type=Instance-based2020.10 | 89.11 | 72.04 | 44.83 | |
| Co-teaching+Noise type=Instance-based2020.10 | 89.04 | 69.15 | 33.33 | |
| LDMINoise type=Instance-based2020.10 | 88.8 | 82.7 | 70.54 | |
| JoCoRNoise type=Instance-based2020.10 | 88.71 | 68.97 | 30.27 | |
| Co-teachingNoise type=Instance-based2020.10 | 88.66 | 69.5 | 34.61 | |
| Forward TNoise type=Instance-based2020.10 | 88.08 | 82.67 | 41.57 | |
| Cross EntropyNoise type=Instance-based2020.10 | 87.16 | 75.16 | 44.64 | |
| LqNoise type=Instance-based2020.10 | 86.45 | 69.02 | 32.94 |