Cross-Domain Class-Incremental Learning on OfficeHome 13 tasks
60.01Accuracy (SD)DER w/ CORF
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DER w/ CORFBackbone=ResNet-342026.05 | 60.01 | 38.54 | 46.17 | |
| DERBackbone=ResNet-342026.05 | 57.39 | 35.09 | 42.41 | |
| MEMO w/ CORFBackbone=ResNet-342026.05 | 55.15 | 33.19 | 40.3 | |
| MEMOBackbone=ResNet-342026.05 | 53.33 | 32.73 | 39.45 | |
| FOSTER w/ CORFBackbone=ResNet-342026.05 | 53.09 | 37.94 | 43.66 | |
| FOSTERBackbone=ResNet-342026.05 | 53.05 | 32.36 | 38.98 | |
| iCaRL w/ CORFBackbone=ResNet-342026.05 | 50.77 | 32.89 | 38.95 | |
| Replay w/ CORFBackbone=ResNet-342026.05 | 50.1 | 31.65 | 37.7 | |
| iCaRLBackbone=ResNet-342026.05 | 49.41 | 29.55 | 35.95 | |
| ReplayBackbone=ResNet-342026.05 | 48.3 | 29.72 | 35.74 | |
| TagFex w/ CORFBackbone=ResNet-342026.05 | 45.98 | 25.2 | 30.86 | |
| DS-AL w/ CORFBackbone=ResNet-342026.05 | 44.62 | 26.23 | 31.15 | |
| DS-ALBackbone=ResNet-342026.05 | 44.26 | 22.02 | 29.84 | |
| TagFexBackbone=ResNet-342026.05 | 43.62 | 22.76 | 28.58 | |
| TRIPSBackbone=ResNet-342026.05 | 31.56 | 17.84 | 20.71 | |
| MSL+MOVBackbone=ResNet-342026.05 | 29.13 | 16.22 | 18.94 | |
| FineTune w/ CORFBackbone=ResNet-342026.05 | 19.15 | 15.11 | 16.49 | |
| FineTuneBackbone=ResNet-342026.05 | 17.2 | 13.31 | 14.77 |