Task Incremental Learning on S-CIFAR-10
98.31AccuracyJOINT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| JOINTBackbone=ResNet-18, Pre-training=None, Batch size=322024.10 | 98.31 | — | |
| CLS-ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 90.75 | — | |
| ER-ACEBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 90.49 | — | |
| CLS-ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 89.71 | — | |
| CLFD-ER-ACEBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 89.35 | — | |
| ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 89.12 | — | |
| DER++Backbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 88.34 | — | |
| CLFD-ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 88.29 | — | |
| CLFD-ER-ACEBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 87.13 | — | |
| CLFD-CLS-ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 87.09 | — | |
| ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 86.36 | — | |
| ER-ACEBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 85.78 | — | |
| CLFD-CLS-ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 85.3 | — | |
| CLFD-DER++Backbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=1252024.10 | 84.76 | — | |
| CLFD-ERBackbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 84.45 | — | |
| DER++Backbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 83.51 | — | |
| CLFD-DER++Backbone=ResNet-18, Pre-training=None, Batch size=32, Buffer size=502024.10 | 81.15 | — | |
| oEWCBackbone=ResNet-18, Pre-training=None, Batch size=322024.10 | 68.29 | — | |
| SIBackbone=ResNet-18, Pre-training=None, Batch size=322024.10 | 68.05 | — | |
| LwFBackbone=ResNet-18, Pre-training=None, Batch size=322024.10 | 63.29 | — | |
| SGDBackbone=ResNet-18, Pre-training=None, Batch size=322024.10 | 61.02 | — |