Few-shot Classification on CIFAR-FS (meta-test) (5-way, Best/Last Accuracy)
41.91Best Accuracy (5-way 1-shot)TDFML-ProtoNet (+mixup)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| TDFML-ProtoNet (+mixup)Base Learner=ProtoNet, Augmentation=mixup2023.11 | 41.91 | 41.78 | 58.25 | 56.12 | |
| TDFML-ProtoNetBase Learner=ProtoNet2023.11 | 40.8 | 40.28 | 57.11 | 55.69 | |
| TDFML-ANILBase Learner=ANIL2023.11 | 40.39 | 39.69 | 55.31 | 52.92 | |
| PURER-ProtoNetBase Learner=ProtoNet2023.11 | 36.26 | 27.01 | 52.67 | 40.53 | |
| PURER-ANILBase Learner=ANIL2023.11 | 35.31 | 26.4 | 51.63 | 41.24 | |
| DFL2L2023.11 | 30.43 | 29.35 | 36.21 | 35.28 | |
| RANDOM2023.11 | 28.59 | 28.59 | 34.77 | 34.77 | |
| AVERAGE2023.11 | 23.96 | 23.96 | 27.04 | 27.04 |