Task-incremental continual learning on Split Mini-ImageNet
17.9AutoAttack AA (%)OGD
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| OGD2025.06 | 17.9 | 18.8 | 20.7 | 29.6 | -0.29 | |
| A-GEM2025.06 | 19 | 19.8 | 21.2 | 29.2 | -0.28 | |
| SI2025.06 | 20.4 | 21.3 | 22.7 | 28.1 | -0.27 | |
| SGD2025.06 | 20.5 | 22 | 23.5 | 30.8 | -0.24 | |
| EWC2025.06 | 21.3 | 22.7 | 24.3 | 29.9 | -0.25 | |
| GEM2025.06 | 22.3 | 23.8 | 25.4 | 31.8 | -0.2 | |
| GPM2025.06 | 26.3 | 27.1 | 28.8 | 36.8 | -0.12 | |
| DGP2025.06 | 32.1 | 33.8 | 35.5 | 44.8 | -0.05 | |
| SHIELD2025.06 | 56.22 | 56.8 | 53.08 | 59.52 | -0.16 | |
| SHIELDIMInterval MixUp=true2025.06 | 57.9 | 58.47 | 54.1 | 62.67 | -0.18 |