Online Class-Incremental Learning on ImageNet-100 10 tasks (test)
56.8FAA (%)MOSE-MOE + GWM
Evaluation Results
| Method | Links | |
|---|---|---|
| MOSE-MOE + GWMMemory Size (Ms)=10K2025.08 | 56.8 | |
| EMI + GWMMemory Size (Ms)=10K2025.08 | 50.9 | |
| ER-DCBA + GWMMemory Size (Ms)=10K2025.08 | 50.5 | |
| GSA + GWMMemory Size (Ms)=10K2025.08 | 49 | |
| HPCR + GWMMemory Size (Ms)=10K2025.08 | 48.4 | |
| SCR + GWMMemory Size (Ms)=10K2025.08 | 46.5 | |
| ER-ACE + GWMMemory Size (Ms)=10K2025.08 | 46.1 | |
| OCM + GWMMemory Size (Ms)=10K2025.08 | 45.1 | |
| HPCR + GWMMemory Size (Ms)=1K, Backbone=ResNet-18, Batch Size=102025.08 | 39.4 | |
| HPCR + GWMMemory Size (Ms)=0.5K, Backbone=ResNet-18, Batch Size=102025.08 | 33.3 | |
| GSA + GWMMemory Size (Ms)=1K, Backbone=ResNet-18, Batch Size=102025.08 | 30.2 | |
| HPCRMemory Size (Ms)=1K, Backbone=ResNet-18, Batch Size=102025.08 | 28 | |
| HPCR + GWMMemory Size (Ms)=0.2K, Backbone=ResNet-18, Batch Size=102025.08 | 26 | |
| GSA + GWMMemory Size (Ms)=0.5K, Backbone=ResNet-18, Batch Size=102025.08 | 23.3 | |
| HPCRMemory Size (Ms)=0.5K, Backbone=ResNet-18, Batch Size=102025.08 | 21.9 | |
| GSAMemory Size (Ms)=1K, Backbone=ResNet-18, Batch Size=102025.08 | 20.2 | |
| HPCRMemory Size (Ms)=0.2K, Backbone=ResNet-18, Batch Size=102025.08 | 16.2 | |
| GSA + GWMMemory Size (Ms)=0.2K, Backbone=ResNet-18, Batch Size=102025.08 | 16.1 | |
| GSAMemory Size (Ms)=0.5K, Backbone=ResNet-18, Batch Size=102025.08 | 15.7 | |
| ERMemory Size (Ms)=2K2025.08 | 14.6 | |
| GSAMemory Size (Ms)=0.2K, Backbone=ResNet-18, Batch Size=102025.08 | 11.5 |