Cross-Domain Class-Incremental Learning on OfficeHome 5 tasks
57.92Average Accuracy (SD)DER w/ CORF
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| DER w/ CORFBackbone=ResNet-342026.05 | 57.92 | 33.82 | 41.65 | |
| FOSTER w/ CORFBackbone=ResNet-342026.05 | 56.71 | 35.28 | 42.49 | |
| FOSTERBackbone=ResNet-342026.05 | 56.08 | 33.15 | 40.68 | |
| MEMO w/ CORFBackbone=ResNet-342026.05 | 56.02 | 33.25 | 40.79 | |
| DERBackbone=ResNet-342026.05 | 55.09 | 32.11 | 39.53 | |
| iCaRL w/ CORFBackbone=ResNet-342026.05 | 54.39 | 33.58 | 40.55 | |
| MEMOBackbone=ResNet-342026.05 | 53.02 | 30.76 | 37.97 | |
| Replay w/ CORFBackbone=ResNet-342026.05 | 51.51 | 31.23 | 38.02 | |
| TagFex w/ CORFBackbone=ResNet-342026.05 | 50.25 | 28.31 | 35.48 | |
| iCaRLBackbone=ResNet-342026.05 | 49.81 | 29.2 | 35.9 | |
| TagFexBackbone=ResNet-342026.05 | 49.81 | 26.13 | 33.97 | |
| DS-AL w/ CORFBackbone=ResNet-342026.05 | 49.35 | 26.47 | 33.64 | |
| ReplayBackbone=ResNet-342026.05 | 48.25 | 28.63 | 35.07 | |
| DS-ALBackbone=ResNet-342026.05 | 47.92 | 22.54 | 28.84 | |
| TRIPSBackbone=ResNet-342026.05 | 36.82 | 17.97 | 23.77 | |
| MSL+MOVBackbone=ResNet-342026.05 | 31.13 | 15.25 | 19.63 | |
| FineTune w/ CORFBackbone=ResNet-342026.05 | 30.9 | 23.33 | 25.86 | |
| FineTuneBackbone=ResNet-342026.05 | 29.24 | 19 | 22.65 |