Few-Shot Class-Incremental Learning on CIFAR100
84.6BCRFACT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FACTBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 84.6 | 52.4 | 68.5 | |
| SAVCBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 84.5 | 47.8 | 66.2 | |
| OrCoBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 78.8 | 31.3 | 55 | |
| Gen1S-VAEBackbone=ResNet18, Conditioning=unconditioned, Shot count=10-shot, Novel classes=12026.01 | 76.4 | 56.7 | 66.6 | |
| LIMITBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 76.1 | 60.5 | 68.3 | |
| SimpleShotBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 76.1 | 49.6 | 62.9 | |
| ProtoNetBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 76 | 47.3 | 61.6 | |
| SLDABackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 75.8 | 43.6 | 59.7 | |
| Gen1S-DiffusionBackbone=ResNet18, Conditioning=unconditioned, Shot count=10-shot, Novel classes=12026.01 | 75 | 48.9 | 62 | |
| Gen1S-VAEBackbone=ResNet18, Conditioning=conditioned, Shot count=10-shot, Novel classes=12026.01 | 74.6 | 55.9 | 65.2 | |
| SLDA-fixed-ΣBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 74.4 | 43.1 | 58.8 | |
| Gen1S-DiffusionBackbone=ResNet18, Conditioning=conditioned, Shot count=10-shot, Novel classes=12026.01 | 74.1 | 51.6 | 62.8 | |
| RelationNetBackbone=ResNet18, Shot count=10-shot, Novel classes=12026.01 | 50.1 | 15.6 | 32.8 |