Multi-label Classification on ChestMNIST
93.6AccuracyEnsemble
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| EnsembleBackbone=ResNet-18, Ensemble members=5, FLOPS=5x2024.06 | 93.6 | 0.053 | 1.162 | |
| Partial BayesianBackbone=ResNet-18, rbayes=1%, FLOPS=3.36x2024.06 | 93.4 | 0.064 | 0.367 | |
| Partial BayesianBackbone=ResNet-18, rbayes=5%, FLOPS=3.45x2024.06 | 93.1 | 0.067 | 0.404 | |
| Partial BayesianBackbone=ResNet-18, rbayes=20%, FLOPS=3.8x2024.06 | 92.5 | 0.083 | 1.303 | |
| DeterministicBackbone=ResNet-18, FLOPS=1x2024.06 | 89.9 | 0.098 | 0.493 | |
| Bayesian NetworkBackbone=ResNet-18, FLOPS=15x2024.06 | 72.3 | 0.215 | 3.794 |