Table 2. Omniglot dataset single sample classification experimental results.
| Model | Fine-tuning | Accuracy | |
|---|---|---|---|
| 5-way 1-shot | 20-way 1-shot | ||
| MAML | Y | 97.80 ± 0.32% | 95.60 ± 0.21% |
| Meta-SGD | Y | 99.50 ± 0.23% | 95.83 ± 0.36% |
| Multi-scale relational networks | N | 99.35 ± 0.25% | 97.41 ± 0.28% |
| Multi-scale meta-relational network | N | 99.57 ± 0.16% | 97.88 ± 0.20% |