Sequential Multi-Task Learning on NYU Sequence 2 shuffled v2 (test)
-27.11Delta T_biCaRL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| iCaRLBackbone=ResNet-182026.03 | -27.11 | 7.67 | |
| DERBackbone=ResNet-182026.03 | -26.55 | 6.67 | |
| EWCBackbone=ResNet-182026.03 | -0.77 | 5.33 | |
| SGPBackbone=ResNet-182026.03 | -0.17 | 6 | |
| Vanilla trainingBackbone=ResNet-182026.03 | 0 | 6 | |
| SPGBackbone=ResNet-182026.03 | 0.29 | 4.67 | |
| LwFBackbone=ResNet-182026.03 | 0.77 | 3.67 | |
| HADBackbone=ResNet-182026.03 | 1.18 | 3.33 | |
| Joint trainingBackbone=ResNet-182026.03 | 3.09 | 1 |