5-way K-shot Classification on mini-ImageNet (test)
46.81Accuracy (5-way, 1-shot)MAML
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MAMLBackbone=Conv-4, Regime=supervised training2020.06 | 46.81 | 62.13 | 71.03 | 75.54 | |
| MAMLArchitecture=Conv-4, Pre-training=Supervised2020.06 | 46.81 | 62.13 | 71.03 | 75.54 | |
| ProtoNetBackbone=Conv-4, Regime=supervised training2020.06 | 46.44 | 66.33 | 76.73 | 78.91 | |
| ProtoNetArchitecture=Conv-4, Pre-training=Supervised2020.06 | 46.44 | 66.33 | 76.73 | 78.91 | |
| ProtoTransferBackbone=Conv-4, Regime=unsupervised2020.06 | 45.67 | 62.99 | 72.34 | 77.22 | |
| ProtoTransferArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 45.67 | 62.99 | 72.34 | 77.22 | |
| Pre+LinearBackbone=Conv-4, Regime=supervised training2020.06 | 43.87 | 63.01 | 75.46 | 80.17 | |
| Pre+LinearArchitecture=Conv-4, Pre-training=Supervised2020.06 | 43.87 | 63.01 | 75.46 | 80.17 | |
| ULDA-MetaOptNetBackbone=Conv-4, Regime=unsupervised2020.06 | 40.71 | 54.49 | 63.58 | 67.65 | |
| ULDA-MetaOptNetArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 40.71 | 54.49 | 63.58 | 67.65 | |
| ULDA-ProtoNetBackbone=Conv-4, Regime=unsupervised2020.06 | 40.63 | 55.41 | 63.16 | 65.2 | |
| ULDA-ProtoNetArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 40.63 | 55.41 | 63.16 | 65.2 | |
| UMTRABackbone=Conv-4, Regime=unsupervised2020.06 | 39.93 | 50.73 | 61.11 | 67.15 | |
| UMTRAArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 39.93 | 50.73 | 61.11 | 67.15 | |
| CACTUS-MAMLBackbone=Conv-4, Regime=unsupervised2020.06 | 39.9 | 53.97 | 63.84 | 69.64 | |
| CACTUS-MAMLArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 39.9 | 53.97 | 63.84 | 69.64 | |
| CACTUS-ProtoNetBackbone=Conv-4, Regime=unsupervised2020.06 | 39.18 | 53.36 | 61.54 | 63.55 | |
| CACTUS-ProtoNetArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 39.18 | 53.36 | 61.54 | 63.55 | |
| AAL-ProtoNetBackbone=Conv-4, Regime=unsupervised2020.06 | 37.67 | 40.29 | — | — | |
| AAL-ProtoNetArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 37.67 | 40.29 | — | — | |
| AAL-MAML++Backbone=Conv-4, Regime=unsupervised2020.06 | 34.57 | 49.18 | — | — | |
| AAL-MAML++Architecture=Conv-4, Pre-training=Unsupervised2020.06 | 34.57 | 49.18 | — | — | |
| UFLSTBackbone=Conv-4, Regime=unsupervised2020.06 | 33.77 | 45.03 | 53.35 | 56.72 | |
| UFLSTArchitecture=Conv-4, Pre-training=Unsupervised2020.06 | 33.77 | 45.03 | 53.35 | 56.72 | |
| Training (scratch)Backbone=Conv-4, Regime=scratch2020.06 | 27.59 | 38.48 | 51.53 | 59.63 | |
| Training (scratch)Architecture=Conv-4, Training=Randomly Initialized2020.06 | 27.59 | 38.48 | 51.53 | 59.63 |