5-way 1-shot Few-shot Image Classification on tieredImageNet (meta-test)
66.33AccuracyLEO
Evaluation results
| Method | Links | |
|---|---|---|
| LEObackbone=WRN-28-102019.04 | 66.33 | |
| MetaOptNet-SVMbackbone=ResNet-122019.04 | 65.99 | |
| MetaOptNet-SVM-trainvalbackbone=ResNet-12, training_set=meta-training + meta-validation2019.04 | 65.81 | |
| MetaOptNet-RRbackbone=ResNet-122019.04 | 65.36 | |
| Transductive Prop Netsbackbone=64-64-64-642019.04 | 59.91 | |
| RFSChannels=642022.01 | 56.8 | |
| HTChannels=642022.01 | 56.3 | |
| HTChannels=482022.01 | 55 | |
| RFSChannels=482022.01 | 54.6 | |
| Relation Networksbackbone=64-96-128-2562019.04 | 54.48 | |
| HTChannels=322022.01 | 54 | |
| Prototypical Networksbackbone=64-64-64-642019.04 | 53.31 | |
| HTChannels=162022.01 | 51.9 | |
| MAMLbackbone=32-32-32-322019.04 | 51.67 | |
| RFSChannels=322022.01 | 51.5 | |
| HTChannels=82022.01 | 49.1 | |
| RFSChannels=162022.01 | 47.7 | |
| RFSChannels=82022.01 | 44.1 |