Medical Image Classification on DermaMNIST 20% Noise
97.7SensitivityUNICON+CS
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| UNICON+CSCS Loss=True2026.04 | 97.7 | 41.1 | 69.4 | 68.7 | 44.4 | |
| Co-teaching+CSCS Loss=True2026.04 | 78.3 | 73.7 | 76 | 82.5 | 54.7 | |
| GMM FilterCS Loss=False2026.04 | 75.8 | 71.1 | 73.4 | 80.9 | 51.4 | |
| GMM Filter+CSCS Loss=True2026.04 | 75.5 | 60.4 | 67.9 | 74.3 | 44.6 | |
| DivideMix+CSCS Loss=True2026.04 | 72.2 | 59.4 | 65.8 | 67.7 | 42.6 | |
| DivideMixCS Loss=False2026.04 | 70.4 | 74.3 | 72.4 | 80.7 | 51 | |
| UNICONCS Loss=False2026.04 | 70.4 | 74.3 | 72.4 | 80.7 | 51 | |
| BaselineCS Loss=False2026.04 | 68.6 | 76.9 | 72.8 | 79.8 | 52.1 | |
| Co-teachingCS Loss=False2026.04 | 61.7 | 85.7 | 73.7 | 84.3 | 56 | |
| Baseline+CSCS Loss=True2026.04 | 59.9 | 71.2 | 65.6 | 72.4 | 43.1 |