Continual Learning on MNIST-R
84.08AccuracyMLP DCL-30-1 MEM
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MLP DCL-30-1 MEMArchitecture=MLP, DCL parameters=30-1, Memory constraints=MEM, Optimizer=SGD2019.12 | 84.08 | 0.0094 | 0.6423 | |
| MLP DCL-40-1 MEMArchitecture=MLP, DCL parameters=40-1, Memory constraints=MEM, Optimizer=SGD2019.12 | 84.02 | 0.0127 | 0.6351 | |
| GEMOptimizer=SGD2019.12 | 83.35 | -0.0047 | 0.6521 | |
| MLP DCL-50-1 MEMArchitecture=MLP, DCL parameters=50-1, Memory constraints=MEM, Optimizer=SGD2019.12 | 82.77 | 0.0238 | 0.6111 | |
| sparse-La-MAML2021.10 | 77.77 | -8.16 | — | |
| MER2021.10 | 77.42 | -5.6 | — | |
| La-M2021.10 | 77.42 | -8.64 | — | |
| GEM2021.10 | 67.38 | -18.02 | — | |
| EWCOptimizer=SGD2019.12 | 54.61 | -0.2087 | 0.5574 | |
| Baseline2021.10 | 53.38 | -5.44 | — |