Class-Incremental Learning on Tiny ImageNet 20 tasks (val)
55.5ANUpper Bound
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Upper BoundReplay Data=None2021.06 | 55.5 | 100 | |
| Always Be Dreaming (Ours)Replay Data=Synthetic2021.06 | 12.1 | 49.3 | |
| LwF.MCReplay Data=None2021.06 | 8.8 | 37.2 | |
| DeepInversionReplay Data=Synthetic2021.06 | 5.1 | 24.8 | |
| LwFReplay Data=None2021.06 | 4.4 | 22.4 | |
| BaseReplay Data=None2021.06 | 4.1 | 21.9 | |
| LwFReplay Data=Synthetic2021.06 | 4 | 22 |