Image Classification Calibration on BloodMNIST
0.3119NLLDirichlet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DirichletK (Number of clusters)=10, λ (Shrinkage parameter)=12025.10 | 0.3119 | 0.1577 | 1.73 | 89.22 | |
| VectorK (Number of clusters)=-, λ (Shrinkage parameter)=-2025.10 | 0.3121 | 0.1573 | 1.72 | 89.35 | |
| clustered Vector scalingK (Number of clusters)=10, λ (Shrinkage parameter)=12025.10 | 0.3121 | 0.1573 | 1.72 | 89.35 | |
| DirichletK (Number of clusters)=-, λ (Shrinkage parameter)=-2025.10 | 0.3123 | 0.1579 | 1.73 | 89.22 | |
| clustered Temperature ScalingK (Number of clusters)=10, λ (Shrinkage parameter)=12025.10 | 0.3152 | 0.1577 | 1.85 | 89.4 | |
| TSK (Number of clusters)=-, λ (Shrinkage parameter)=-2025.10 | 0.3156 | 0.1579 | 1.88 | 89.4 | |
| BaseK (Number of clusters)=-, λ (Shrinkage parameter)=-2025.10 | 0.3185 | 0.1593 | 2.69 | 89.4 | |
| Binwise-TSK (Number of clusters)=-, λ (Shrinkage parameter)=-2025.10 | 0.3305 | 0.1574 | 1.41 | 89.4 | |
| Binwise-VectorK (Number of clusters)=-, λ (Shrinkage parameter)=-2025.10 | 0.3334 | 0.1605 | 1.77 | 89.25 |