Online Task-incremental Learning on Split CIFAR100
71AccuracyLayerwise Optimization by Gradient Decomposition
Evaluation Results
| Method | Links | |
|---|---|---|
| Layerwise Optimization by Gradient DecompositionTraining Epochs=3 epochs2021.05 | 71 | |
| GEMTraining Epochs=3 epochs, Memory Size=2562021.05 | 69.6 | |
| Layerwise Optimization by Gradient DecompositionTraining Epochs=1 epoch2021.05 | 69.3 | |
| A-GEMTraining Epochs=3 epochs, Memory Size=2562021.05 | 69.1 | |
| S-GEMTraining Epochs=3 epochs, Memory Size=2562021.05 | 68.8 | |
| tinyER+Training Epochs=1 epoch2021.05 | 68.3 | |
| GEMTraining Epochs=1 epoch, Memory Size=2562021.05 | 65.8 | |
| A-GEMTraining Epochs=1 epoch, Memory Size=2562021.05 | 65.3 | |
| S-GEMTraining Epochs=1 epoch, Memory Size=2562021.05 | 64.2 | |
| GDumbTraining Epochs=1 epoch, Training Schedule=Online Task Incremental Disjoint2021.05 | 60.3 | |
| EWCTraining Epochs=3 epochs2021.05 | 56.9 | |
| iCARLTraining Epochs=3 epochs2021.05 | 55.8 | |
| iCARLTraining Epochs=1 epoch2021.05 | 55.7 | |
| SingleTraining Epochs=3 epochs2021.05 | 55.4 | |
| SingleTraining Epochs=1 epoch2021.05 | 54.7 | |
| IndependentTraining Epochs=3 epochs2021.05 | 53.5 | |
| EWCTraining Epochs=1 epoch2021.05 | 48.3 | |
| IndependentTraining Epochs=1 epoch2021.05 | 43.3 |