Class-incremental Learning on CIFAR100 (B=5, C=5) (test)
78Avg AccOracle w/ LingoCL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Oracle w/ LingoCLBackbone=ResNet-182024.03 | 78 | 78 | — | |
| OracleBackbone=ResNet-182024.03 | 77.6 | 77.6 | — | |
| DyTox w/ LingoCLBackbone=ConViT2024.03 | 70 | 52.3 | 30.5 | |
| DyToxBackbone=ConViT2024.03 | 67.4 | 48.1 | 37.8 | |
| AANet w/ LingoCLBackbone=ResNet-182024.03 | 63.2 | 44.5 | — | |
| AANetBackbone=ResNet-182024.03 | 62.5 | 42.5 | — | |
| IL2M w/ LingoCLBackbone=ResNet-182024.03 | 56.6 | 44 | 43 | |
| LUCIR w/ LingoCLBackbone=ResNet-182024.03 | 56.3 | 44.3 | 39.8 | |
| CwD w/ LingoCLBackbone=ResNet-182024.03 | 55.7 | 44.3 | 39.2 | |
| LUCIRBackbone=ResNet-182024.03 | 54.8 | 41.7 | 42 | |
| CwDBackbone=ResNet-182024.03 | 54.4 | 42.2 | 41.5 | |
| BiMeCo w/ LingoCLBackbone=MobileNet-V2†2024.03 | 54.2 | 39.1 | 41.8 | |
| IL2MBackbone=ResNet-182024.03 | 52.6 | 40.5 | 45.3 | |
| BiMeCoBackbone=MobileNet-V2†2024.03 | 52.5 | 38.7 | 43.5 | |
| BiC w/ LingoCLBackbone=ResNet-182024.03 | 51.6 | 36.6 | 19.6 | |
| BiCBackbone=ResNet-182024.03 | 50.1 | 34.7 | 28.7 |