Class-Incremental Image Classification on ImageNet 100
18.2Last-task Forgetting (Flast)BiCyc
Evaluation results
| Method | Links | |
|---|---|---|
| BiCycT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 18.2 | |
| AdaGaussT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 20.6 | |
| BiCycT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 20.8 | |
| EFCT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 21.5 | |
| AdaGaussT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 22.9 | |
| EFCT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 23.8 | |
| LDCT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 25.7 | |
| ADCT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 32.4 | |
| LDCT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 32.9 | |
| ADCT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 33.4 | |
| LwFT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 42.1 | |
| SDCT (Number of tasks)=10, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 44.6 | |
| LwFT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 48.1 | |
| SDCT (Number of tasks)=20, Backbone=ResNet-18, Batch size=256, Training Initialization=from-scratch2026.06 | 54.4 |