Continual Learning on Split-MNIST
99.98Mean Test AccuracyIsolated
Evaluation results
| Method | Links | |
|---|---|---|
| IsolatedReplay=false, Single Epoch=false2021.06 | 99.98 | |
| Multi-Head (multi-task)2021.06 | 99.98 | |
| Model ZooReplay=true, Single Epoch=false2021.06 | 99.97 | |
| RMN2021.02 | 99.5 | |
| RMNReplay=false, Single Epoch=false2021.06 | 99.5 | |
| HATPretrained network=true, Multiheaded=true2021.02 | 99 | |
| FROMPData replay buffer=true, Points per task=402021.02 | 99 | |
| FROMPReplay=true, Single Epoch=false2021.06 | 99 | |
| HATReplay=false, Single Epoch=false2021.06 | 99 | |
| SIReplay=false, Single Epoch=false2021.06 | 98.9 | |
| OGDReplay=false, Single Epoch=true2021.06 | 98.84 | |
| VCLReplay=true, Single Epoch=false2021.06 | 98.4 | |
| FRCLData replay buffer=true, Points per task=402021.02 | 97.8 | |
| FRCLReplay=true, Single Epoch=false2021.06 | 97.8 | |
| VCLData replay buffer=true, Multiheaded=true, Points per task=402021.02 | 97 | |
| RWALKData replay buffer=true, Multiheaded=true2021.02 | 82.5 | |
| EWC2021.02 | 63.1 | |
| DLP2021.02 | 61.2 | |
| SI2021.02 | 57.6 |