Semi-Supervised Few-Shot Classification on miniImagenet (test)
79.221-shot AccuracyEPNetSSL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EPNetSSLBackbone=WRN-28-10, Number of unlabeled samples=1002020.03 | 79.22 | 88.05 | |
| --NetSSLBackbone=WRN-28-10, Number of unlabeled samples=1002020.03 | 77.7 | 86.3 | |
| EPNetSSLBackbone=RESNET-12, Number of unlabeled samples=1002020.03 | 75.36 | 84.07 | |
| --NetSSLBackbone=RESNET-12, Number of unlabeled samples=1002020.03 | 73.42 | 83.17 | |
| EPNetBackbone=WRN-28-10, Number of unlabeled samples=1002020.03 | 70.74 | 84.34 | |
| LSTBackbone=RESNET-12, Number of unlabeled samples=1002020.03 | 70.1 | 78.7 | |
| EPNetBackbone=RESNET-12, Number of unlabeled samples=1002020.03 | 66.5 | 81.06 | |
| EPNetSSLBackbone=CONV-4, Number of unlabeled samples=1002020.03 | 65.13 | 75.42 | |
| --NetSSLBackbone=CONV-4, Number of unlabeled samples=1002020.03 | 63.74 | 75.3 | |
| TransMatchBackbone=WRN-28-10, Number of unlabeled samples=1002020.03 | 63.02 | 81.19 | |
| EPNetBackbone=CONV-4, Number of unlabeled samples=1002020.03 | 59.32 | 72.95 | |
| TPNSSLBackbone=CONV-4, Number of unlabeled samples=1002020.03 | 52.78 | 66.42 | |
| k-Meansmasked,softBackbone=WRN-28-10, Number of unlabeled samples=1002020.03 | 52.78 | 66.42 | |
| k-Meansmasked,softBackbone=CONV-4, Number of unlabeled samples=1002020.03 | 50.41 | 64.39 |