Class-Incremental Learning on DomainNet 5-task
60.4Average Accuracy (SD)FOSTER w/ CORF
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FOSTER w/ CORFBackbone=ResNet-342026.05 | 60.4 | — | 40.6 | 47.16 | |
| FOSTERBackbone=ResNet-342026.05 | 58.3 | — | 36.3 | 42.54 | |
| DER w/ CORFBackbone=ResNet-342026.05 | 53.36 | — | 42.95 | 47.24 | |
| MEMO w/ CORFBackbone=ResNet-342026.05 | 51.17 | — | 43.47 | 46.7 | |
| DERBackbone=ResNet-342026.05 | 50.66 | — | 30.97 | 36.37 | |
| iCaRL w/ CORFBackbone=ResNet-342026.05 | 47.83 | — | 32.39 | 36.99 | |
| MEMOBackbone=ResNet-342026.05 | 46.73 | — | 28.44 | 33.56 | |
| iCaRLBackbone=ResNet-342026.05 | 45.98 | — | 28.06 | 33.22 | |
| Replay w/ CORFBackbone=ResNet-342026.05 | 44.56 | — | 30.98 | 35.48 | |
| ReplayBackbone=ResNet-342026.05 | 41.79 | — | 26.12 | 30.72 | |
| TagFex w/ CORFBackbone=ResNet-342026.05 | 37.48 | — | 31.93 | 34.87 | |
| TagFexBackbone=ResNet-342026.05 | 35.41 | — | 26.23 | 30.1 | |
| FineTune w/ CORFBackbone=ResNet-342026.05 | 34.99 | — | 26.26 | 28.96 | |
| FineTuneBackbone=ResNet-342026.05 | 34.53 | — | 22.9 | 26.47 | |
| DS-AL w/ CORFBackbone=ResNet-342026.05 | 34.23 | — | 21.31 | 26.85 | |
| DS-ALBackbone=ResNet-342026.05 | 32.91 | — | 19.53 | 21.6 | |
| TRIPSBackbone=ResNet-342026.05 | 24.38 | — | 15.26 | 16.94 | |
| MSL+MOVBackbone=ResNet-342026.05 | 21.04 | — | 16.33 | 17.02 | |
| CODA-PromptTask Scale=5-task2026.03 | — | 73.24 | — | — | |
| DualPromptTask Scale=5-task2026.03 | — | 70.73 | — | — | |
| HiDe-PromptTask Scale=5-task2026.03 | — | 78.92 | — | — | |
| L2PTask Scale=5-task2026.03 | — | 69.58 | — | — | |
| S-Prompt++Task Scale=5-task2026.03 | — | 72.35 | — | — | |
| SAMoETask Scale=5-task2026.03 | — | 78.65 | — | — |