5-way Few-shot Classification on Caltech-256
73.21-shot AccuracyΔ-encoder
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Δ-encoderBackbone features=ResNet2018.06 | 73.2 | 83.6 | |
| Dual TriNetLabel embedding=true2018.06 | 63.8 | 80.5 | |
| DEML+Meta-SGDExternal training data=true2018.06 | 62.2 | 79.5 | |
| Nearest neighbor (baseline)2018.06 | 51.3 | 67.5 | |
| Matching Nets2018.06 | 48.1 | 57.5 | |
| MAML2018.06 | 45.6 | 54.6 |