Few-Shot Class-Incremental Learning on CIFAR100 Single-class five-shot, 1 novel class
84.7BCRFACT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FACTBackbone=ResNet182026.01 | 84.7 | 44.8 | 64.8 | |
| SAVCBackbone=ResNet182026.01 | 84.6 | 52.8 | 68.7 | |
| OrCoBackbone=ResNet182026.01 | 78.8 | 28.1 | 53.4 | |
| SimpleShotBackbone=ResNet182026.01 | 76.4 | 45.3 | 60.9 | |
| Gen1S-VAEBackbone=ResNet18, Conditioning=unconditioned2026.01 | 76.4 | 39.7 | 58.1 | |
| ProtoNetBackbone=ResNet182026.01 | 76.1 | 46.6 | 61.4 | |
| LIMITBackbone=ResNet182026.01 | 76.1 | 42.2 | 59.2 | |
| SLDABackbone=ResNet182026.01 | 75.8 | 17.2 | 46.5 | |
| SLDA-fixed-ΣBackbone=ResNet182026.01 | 75.5 | 18.9 | 47.2 | |
| Gen1S-DiffusionBackbone=ResNet18, Conditioning=unconditioned2026.01 | 75.1 | 42.7 | 58.9 | |
| Gen1S-VAEBackbone=ResNet18, Conditioning=conditioned2026.01 | 74.7 | 49.1 | 61.9 | |
| Gen1S-DiffusionBackbone=ResNet18, Conditioning=conditioned2026.01 | 74.1 | 43.2 | 58.6 | |
| RelationNetBackbone=ResNet182026.01 | 50.1 | 7.6 | 28.8 |