Object Classification on miniImageNet (test)
92.06AccuracyOracle-MAML (control)
Evaluation results
| Method | Links | |
|---|---|---|
| Oracle-MAML (control)way=5, shot=50, ImageNet-scale meta-training=true2018.10 | 92.06 | |
| Oracle-MAML (control)way=5, shot=20, ImageNet-scale meta-training=true2018.10 | 91.14 | |
| Embedding linear classifier (DeepCluster)way=5, shot=50, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 90.1 | |
| Embedding linear classifier (DeepCluster)way=5, shot=20, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 87.41 | |
| Oracle-MAML (control)way=5, shot=5, ImageNet-scale meta-training=true2018.10 | 86.23 | |
| Semi-supervised MAMLway=5, shot=50, ImageNet-scale meta-training=true2018.10 | 85.27 | |
| CACTUS-MAMLway=5, shot=50, P (number of partitions)=1, k (number of clusters)=10000, ImageNet-scale meta-training=true2018.10 | 85.25 | |
| Embedding k-nearest neighbors (DeepCluster)way=5, shot=50, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 84.72 | |
| CACTUS-MAMLway=5, shot=20, P (number of partitions)=1, k (number of clusters)=10000, ImageNet-scale meta-training=true2018.10 | 82.85 | |
| Semi-supervised MAMLway=5, shot=20, ImageNet-scale meta-training=true2018.10 | 82.83 | |
| Embedding k-nearest neighbors (DeepCluster)way=5, shot=20, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 79.81 | |
| Embedding linear classifier (DeepCluster)way=5, shot=5, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 79.05 | |
| Semi-supervised MAMLway=5, shot=5, ImageNet-scale meta-training=true2018.10 | 76.43 | |
| CACTUS-MAMLway=5, shot=5, P (number of partitions)=1, k (number of clusters)=10000, ImageNet-scale meta-training=true2018.10 | 76.42 | |
| Oracle-MAML (control)way=5, shot=1, ImageNet-scale meta-training=true2018.10 | 74.52 | |
| Embedding k-nearest neighbors (DeepCluster)way=5, shot=5, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 69.34 | |
| Semi-supervised MAMLway=5, shot=1, ImageNet-scale meta-training=true2018.10 | 61.75 | |
| CACTUS-MAMLway=5, shot=1, P (number of partitions)=1, k (number of clusters)=10000, ImageNet-scale meta-training=true2018.10 | 60.11 | |
| Training from scratchway=5, shot=50, ImageNet-scale meta-training=true2018.10 | 59.68 | |
| Embedding linear classifier (DeepCluster)way=5, shot=1, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 58.73 | |
| Training from scratchway=5, shot=20, ImageNet-scale meta-training=true2018.10 | 53.67 | |
| Embedding k-nearest neighbors (DeepCluster)way=5, shot=1, d (dimensionality of embedding)=256, ImageNet-scale meta-training=true2018.10 | 50.17 | |
| Training from scratchway=5, shot=5, ImageNet-scale meta-training=true2018.10 | 39.71 | |
| Training from scratchway=5, shot=1, ImageNet-scale meta-training=true2018.10 | 28.64 |