Few-shot classification on Omniglot (test)
99.58AccuracyProtoNets
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ProtoNetsway=5, shot=5, Learning Paradigm=Supervised meta-learning2023.03 | 99.58 | — | — | — | — | |
| Units-MLWay=20-way, Shot=5-shot, Uncertainty Threshold=0.12022.03 | 99.42 | — | — | — | — | |
| ProtoNetN=5, K=5, supervision=supervised2022.02 | 99.28 | — | — | — | — | |
| UFLSTN=5, K=52022.02 | 99.19 | — | — | — | — | |
| PsCoway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 99.13 | — | — | — | — | |
| PsCoway=5, shot=52023.03 | 99.13 | — | — | — | — | |
| Pre+LinearN=5, K=5, supervision=supervised2022.02 | 99.08 | — | — | — | — | |
| Units-MLWay=20-way, Shot=5-shot, Uncertainty Threshold=0.22022.03 | 99.04 | — | — | — | — | |
| MAMLWay=20-way, Shot=5-shot2022.03 | 98.9 | — | — | — | — | |
| Units-MLWay=20-way, Shot=1-shot, Uncertainty Threshold=0.12022.03 | 98.85 | — | — | — | — | |
| MAMLN=5, K=5, supervision=supervised2022.02 | 98.83 | — | — | — | — | |
| MAMLway=5, shot=5, Learning Paradigm=Supervised meta-learning2023.03 | 98.83 | — | — | — | — | |
| ProtoNetsway=20, shot=5, Learning Paradigm=Supervised meta-learning2023.03 | 98.81 | — | — | — | — | |
| VERSAWay=20-way, Shot=5-shot2022.03 | 98.77 | — | — | — | — | |
| ProtoNetsway=5, shot=1, Learning Paradigm=Supervised meta-learning2023.03 | 98.35 | — | — | — | — | |
| AAL-MAML++N=5, K=52022.02 | 97.96 | — | — | — | — | |
| Units-MLWay=20-way, Shot=5-shot2022.03 | 97.72 | — | — | — | — | |
| ProtoNetN=5, K=1, supervision=supervised2022.02 | 97.7 | — | — | — | — | |
| MMCN=5, K=1, supervision=supervised2022.02 | 97.68 | — | — | — | — | |
| VERSAWay=20-way, Shot=1-shot2022.03 | 97.66 | — | — | — | — | |
| C3LRN=5, K=52022.02 | 97.38 | — | — | — | — | |
| Units-NTSWay=20-way, Shot=5-shot, Task Selection=None2022.03 | 97.38 | — | — | — | — | |
| MoCo v2way=5, shot=5, Learning Paradigm=Self-supervised learning2023.03 | 97.38 | — | — | — | — | |
| MoCo v2way=5, shot=52023.03 | 97.38 | — | — | — | — | |
| SwAVway=5, shot=5, Learning Paradigm=Self-supervised learning2023.03 | 97.32 | — | — | — | — | |
| SwAVway=5, shot=52023.03 | 97.32 | — | — | — | — | |
| Meta-SVEBMway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 97.21 | — | — | — | — | |
| ReptileWay=20-way, Shot=5-shot2022.03 | 97.12 | — | — | — | — | |
| Meta-GMVAEway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 97.09 | — | — | — | — | |
| PsCoway=20, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 97.07 | — | — | — | — | |
| PsCoway=20, shot=52023.03 | 97.07 | — | — | — | — | |
| SimCLRway=5, shot=5, Learning Paradigm=Self-supervised learning2023.03 | 97.06 | — | — | — | — | |
| SimCLRway=5, shot=52023.03 | 97.06 | — | — | — | — | |
| UFLSTN=5, K=12022.02 | 97.03 | — | — | — | — | |
| Units-MLWay=20-way, Shot=1-shot, Uncertainty Threshold=0.22022.03 | 96.83 | — | — | — | — | |
| ProtoTransferN=5, K=5, L=2002022.02 | 96.54 | — | — | — | — | |
| ProtoTransferN=5, K=5, L=502022.02 | 96.48 | — | — | — | — | |
| PsCoway=5, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 96.37 | — | — | — | — | |
| PsCoway=5, shot=12023.03 | 96.37 | — | — | — | — | |
| MAMLway=20, shot=5, Learning Paradigm=Supervised meta-learning2023.03 | 96.29 | — | — | — | — | |
| MAMLWay=20-way, Shot=1-shot2022.03 | 95.8 | — | — | — | — | |
| UMTRAN=5, K=52022.02 | 95.43 | — | — | — | — | |
| UMTRAway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 95.43 | — | — | — | — | |
| ProtoNetsway=20, shot=1, Learning Paradigm=Supervised meta-learning2023.03 | 95.31 | — | — | — | — | |
| LASIUMN=5, K=52022.02 | 95.29 | — | — | — | — | |
| LASIUM-MAMLway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 95.29 | — | — | — | — | |
| Meta-GMVAEway=5, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 94.92 | — | — | — | — | |
| MAMLN=5, K=1, supervision=supervised2022.02 | 94.46 | — | — | — | — | |
| MAMLway=5, shot=1, Learning Paradigm=Supervised meta-learning2023.03 | 94.46 | — | — | — | — | |
| Pre+LinearN=5, K=1, supervision=supervised2022.02 | 94.3 | — | — | — | — | |
| Units-MLWay=20-way, Shot=1-shot2022.03 | 93.17 | — | — | — | — | |
| SwAVway=5, shot=1, Learning Paradigm=Self-supervised learning2023.03 | 93.13 | — | — | — | — | |
| SwAVway=5, shot=12023.03 | 93.13 | — | — | — | — | |
| MoCo v2way=5, shot=1, Learning Paradigm=Self-supervised learning2023.03 | 92.66 | — | — | — | — | |
| MoCo v2way=5, shot=12023.03 | 92.66 | — | — | — | — | |
| MoCo v2way=20, shot=5, Learning Paradigm=Self-supervised learning2023.03 | 92.35 | — | — | — | — | |
| MoCo v2way=20, shot=52023.03 | 92.34 | — | — | — | — | |
| Meta-SVEBMway=20, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 92.21 | — | — | — | — | |
| SimCLRway=5, shot=1, Learning Paradigm=Self-supervised learning2023.03 | 92.13 | — | — | — | — | |
| SimCLRway=5, shot=12023.03 | 92.13 | — | — | — | — | |
| UMTRAway=20, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 92.12 | — | — | — | — | |
| SwAVway=20, shot=5, Learning Paradigm=Self-supervised learning2023.03 | 92.12 | — | — | — | — | |
| SwAVway=20, shot=52023.03 | 92.12 | — | — | — | — | |
| Units-NTSWay=20-way, Shot=1-shot, Task Selection=None2022.03 | 91.96 | — | — | — | — | |
| Meta-SVEBMway=5, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 91.85 | — | — | — | — | |
| SimCLRway=20, shot=5, Learning Paradigm=Self-supervised learning2023.03 | 91.6 | — | — | — | — | |
| SimCLRway=20, shot=52023.03 | 91.6 | — | — | — | — | |
| LASIUM-ProtoNetsway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 91.1 | — | — | — | — | |
| Meta-GMVAEway=20, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 90.61 | — | — | — | — | |
| PsCoway=20, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 89.64 | — | — | — | — | |
| PsCoway=20, shot=12023.03 | 89.6 | — | — | — | — | |
| ReptileWay=20-way, Shot=1-shot2022.03 | 89.43 | — | — | — | — | |
| C3LRN=5, K=12022.02 | 89.3 | — | — | — | — | |
| AAL-ProtoNetN=5, K=52022.02 | 89.14 | — | — | — | — | |
| AAL-MAML++N=5, K=12022.02 | 88.4 | — | — | — | — | |
| ProtoTransferN=5, K=1, L=2002022.02 | 88.37 | — | — | — | — | |
| ProtoTransferN=5, K=1, L=502022.02 | 88 | — | — | — | — | |
| CACTUS-MAMLN=5, K=52022.02 | 87.78 | — | — | — | — | |
| CACTUS-MAMLway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 87.78 | — | — | — | — | |
| AAL-ProtoNetN=5, K=12022.02 | 84.66 | — | — | — | — | |
| MAMLway=20, shot=1, Learning Paradigm=Supervised meta-learning2023.03 | 84.6 | — | — | — | — | |
| UMTRAN=5, K=12022.02 | 83.8 | — | — | — | — | |
| UMTRAway=5, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 83.8 | — | — | — | — | |
| CACTUS-ProtoNetN=5, K=52022.02 | 83.58 | — | — | — | — | |
| CACTUS-ProtoNetsway=5, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 83.58 | — | — | — | — | |
| LASIUMN=5, K=12022.02 | 83.26 | — | — | — | — | |
| LASIUM-MAMLway=5, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 83.26 | — | — | — | — | |
| SwAVway=20, shot=1, Learning Paradigm=Self-supervised learning2023.03 | 82.63 | — | — | — | — | |
| SwAVway=20, shot=12023.03 | 82.63 | — | — | — | — | |
| Meta-GMVAEway=20, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 82.21 | — | — | — | — | |
| MoCo v2way=20, shot=1, Learning Paradigm=Self-supervised learning2023.03 | 82.13 | — | — | — | — | |
| MoCo v2way=20, shot=12023.03 | 82.13 | — | — | — | — | |
| SimCLRway=20, shot=1, Learning Paradigm=Self-supervised learning2023.03 | 80.95 | — | — | — | — | |
| SimCLRway=20, shot=12023.03 | 80.95 | — | — | — | — | |
| LASIUM-ProtoNetsway=5, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 80.15 | — | — | — | — | |
| Meta-SVEBMway=20, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 79.66 | — | — | — | — | |
| Training from Scratchway=5, shot=5, Learning Paradigm=Supervised learning2023.03 | 74.78 | — | — | — | — | |
| UMTRAway=20, shot=1, Learning Paradigm=Unsupervised meta-learning2023.03 | 74.25 | — | — | — | — | |
| CACTUS-MAMLway=20, shot=5, Learning Paradigm=Unsupervised meta-learning2023.03 | 73.36 | — | — | — | — | |
| CACTUS-MAMLN=5, K=12022.02 | 68.84 | — | — | — | — |