Medical Image Classification on PathMNIST 0% Noise
88.7SensitivityDivideMix+CS
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DivideMix+CSCS Loss=True2026.04 | 88.7 | 90.5 | 89.6 | 95.2 | 80.5 | |
| UNICON+CSCS Loss=True2026.04 | 88.7 | 90.5 | 89.6 | 95.2 | 80.5 | |
| Co-teachingCS Loss=False2026.04 | 84.3 | 98.4 | 91.3 | 97.3 | 88.8 | |
| Co-teaching+CSCS Loss=True2026.04 | 83.5 | 97.1 | 90.3 | 96.1 | 86.4 | |
| GMM Filter+CSCS Loss=True2026.04 | 83.2 | 97.1 | 90.2 | 96.1 | 86.3 | |
| DivideMixCS Loss=False2026.04 | 82.9 | 96.1 | 89.5 | 96.9 | 46.3 | |
| UNICONCS Loss=False2026.04 | 82.4 | 97.2 | 89.8 | 96.9 | 85.9 | |
| GMM FilterCS Loss=False2026.04 | 82.3 | 97.9 | 90.1 | 96.3 | 87 | |
| BaselineCS Loss=False2026.04 | 82 | 98.1 | 90.1 | 96.1 | 87 | |
| Baseline+CSCS Loss=True2026.04 | 81.4 | 98 | 89.7 | 96.2 | 65 |