Multiple Instance Learning classification on Eight Pathology Tasks (held-out)
80.1Weighted F1 ScoreMOOZY
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MOOZYAggregation=Average over 5 MIL architectures (MeanMIL, ABMIL, CLAM, DSMIL, TransMIL)2026.03 | 80.1 | 81.5 | 75.8 | |
| CONCH v1.5Aggregation=Average over 5 MIL architectures (MeanMIL, ABMIL, CLAM, DSMIL, TransMIL)2026.03 | 74.6 | 75.1 | 69.6 | |
| BackboneAggregation=Average over 5 MIL architectures (MeanMIL, ABMIL, CLAM, DSMIL, TransMIL)2026.03 | 73.3 | 73.5 | 68.6 | |
| MUSKAggregation=Average over 5 MIL architectures (MeanMIL, ABMIL, CLAM, DSMIL, TransMIL)2026.03 | 72.9 | 72.5 | 67.9 | |
| UNI v2Aggregation=Average over 5 MIL architectures (MeanMIL, ABMIL, CLAM, DSMIL, TransMIL)2026.03 | 71.6 | 71.9 | 66 | |
| Phikon v2Aggregation=Average over 5 MIL architectures (MeanMIL, ABMIL, CLAM, DSMIL, TransMIL)2026.03 | 71.5 | 72.4 | 65.4 |