Medical Image Classification on PathMNIST 20% Noise
95SensitivityUNICON+CS
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| UNICON+CSCS Loss=True2026.04 | 95 | 61.7 | 75.7 | 79.1 | 80.5 | |
| Co-teaching+CSCS Loss=True2026.04 | 86.5 | 93.3 | 89.9 | 96 | 82.8 | |
| Co-teachingCS Loss=False2026.04 | 85.7 | 95.4 | 90.6 | 97.1 | 85.3 | |
| GMM FilterCS Loss=False2026.04 | 85 | 92.2 | 88.6 | 95.4 | 80.6 | |
| GMM Filter+CSCS Loss=True2026.04 | 84.5 | 89.2 | 86.8 | 94.3 | 76.6 | |
| DivideMixCS Loss=False2026.04 | 83.3 | 93.3 | 88.3 | 94.9 | 81 | |
| BaselineCS Loss=False2026.04 | 81.6 | 95.7 | 88.6 | 89.8 | 83.3 | |
| Baseline+CSCS Loss=True2026.04 | 75.8 | 70.7 | 73.2 | 82.2 | 55.4 | |
| UNICONCS Loss=False2026.04 | 73.2 | 95.2 | 84.2 | 94.2 | 77.4 | |
| DivideMix+CSCS Loss=True2026.04 | 62.8 | 82.6 | 72.7 | 79.7 | 56.9 |