Class Incremental Learning on CIFAR100 Ns=25 (test)
0.8041AccuracyJoint (ResNet18)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Joint (ResNet18)Backbone=ResNet182023.03 | 0.8041 | 0 | 0 | — | |
| Joint (Transf.)Backbone=Vision Transformer2023.03 | 0.7612 | 0 | 0 | — | |
| DNEBackbone=Vision Transformer, Heads per task=42023.03 | 0.7147 | 0.757 | 4.65 | — | |
| FOSTERBackbone=ResNet182023.03 | 0.7014 | 0.7633 | 10.27 | — | |
| DytoxBackbone=Vision Transformer2023.03 | 0.6929 | 0.741 | 6.83 | — | |
| DERBackbone=ResNet182023.03 | 0.6908 | 0.7482 | 11.33 | — | |
| DNEBackbone=Vision Transformer, Heads per task=22023.03 | 0.6899 | 0.7391 | 7.13 | — | |
| DNEBackbone=Vision Transformer, Heads per task=12023.03 | 0.6761 | 0.7327 | 8.51 | — | |
| PODNetBackbone=ResNet182023.03 | 0.6154 | 0.7145 | 18.87 | — | |
| iCaRLBackbone=ResNet182023.03 | 0.5379 | 0.668 | 26.62 | — |