Few-shot classification on Omniglot standard (Random Split)
99.7Accuracy (5-way 1-shot)VERSA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| VERSABackbone=Non-standard CNNs2019.07 | 99.7 | 99.75 | 97.66 | 98.77 | |
| RELATION NETSBackbone=Non-standard CNNs2019.07 | 99.6 | 99.8 | 97.6 | 99.1 | |
| PROTO. NETSBackbone=Standard 4-layer CNN, Training Episode=60-way2019.07 | 98.8 | 99.7 | 96 | 98.9 | |
| MAMLBackbone=Standard 4-layer CNN2019.07 | 98.7 | 99.9 | 95.8 | 98.9 | |
| VAMPIREBackbone=Standard 4-layer CNN2019.07 | 98.43 | 99.56 | 93.2 | 98.52 | |
| MEMORY MODULEBackbone=Non-standard CNNs2019.07 | 98.4 | 99.6 | 95 | 98.6 | |
| MATCHING NETSBackbone=Standard 4-layer CNN2019.07 | 98.1 | 98.9 | 93.8 | 98.5 | |
| NEURAL STATISTICIANBackbone=Non-standard CNNs2019.07 | 98.1 | 99.5 | 93.2 | 98.1 | |
| SIAMESE NETSBackbone=Non-standard CNNs2019.07 | 97.3 | 98.4 | 88.2 | 97 |