Few-shot Image Classification on CIFAR-10 Laplace5 noise 1.0 (test)
50.58AccuracyPG-DRO
Evaluation results
| Method | Links | |
|---|---|---|
| PG-DROk (shots)=10, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 50.58 | |
| W-DROk (shots)=10, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 50.21 | |
| SAAk (shots)=10, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 49.68 | |
| PG-DROk (shots)=5, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 41.53 | |
| W-DROk (shots)=5, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 40.49 | |
| SAAk (shots)=5, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 39.42 | |
| Few-shotk (shots)=10, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 32.24 | |
| Few-shotk (shots)=5, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 25.58 | |
| PG-DROk (shots)=1, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 17.71 | |
| W-DROk (shots)=1, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 15.96 | |
| SAAk (shots)=1, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 15.25 | |
| Few-shotk (shots)=1, Backbone=ResNet-18, Pre-training Dataset=CIFAR-1002026.02 | 10.05 |