Few-shot classification on CUB cross-domain from tiered-ImageNet
75.97AccuracyProtoNet + SiMT
Evaluation Results
| Method | Links | |
|---|---|---|
| ProtoNet + SiMTShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 75.97 | |
| ProtoNetShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 74.35 | |
| MetaSGD + SiMTShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 73.93 | |
| MetaSGDShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 70.38 | |
| MAML + SiMTShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 67.58 | |
| ANIL + SiMTShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 66.3 | |
| MAMLShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 65 | |
| ANILShots=5-shot, Ways=5-way, Backbone=ResNet-122022.10 | 62.48 | |
| ProtoNet + SiMTShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 48.89 | |
| ProtoNetShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 47.75 | |
| MetaSGD + SiMTShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 46.62 | |
| MetaSGDShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 46.48 | |
| ANIL + SiMTShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 44.53 | |
| MAML + SiMTShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 44.33 | |
| MAMLShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 42.32 | |
| ANILShots=1-shot, Ways=5-way, Backbone=ResNet-122022.10 | 42.29 |