Class-Incremental Learning on ImageNet-1K standard (test val)
76.42Average AccuracyD³Former Joint
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| D³Former JointNumber of tasks (N)=52022.07 | 76.42 | 76.42 | — | |
| D³Former JointNumber of tasks (N)=102022.07 | 76.42 | 76.42 | — | |
| DyTox JointNumber of tasks (N)=52022.07 | 73.58 | 73.58 | — | |
| DyTox JointNumber of tasks (N)=102022.07 | 73.58 | 73.58 | — | |
| D³FormerNumber of tasks (N)=52022.07 | 72.73 | 64.58 | 21.41 | |
| D³Former-NCMNumber of tasks (N)=52022.07 | 72.61 | 64.64 | 17.03 | |
| D³Former-NCMNumber of tasks (N)=102022.07 | 70.04 | 59.9 | 27.87 | |
| D³FormerNumber of tasks (N)=102022.07 | 69.56 | 59.22 | 32.35 | |
| DyToxNumber of tasks (N)=52022.07 | 68.96 | 64.08 | 18.63 | |
| DyToxNumber of tasks (N)=102022.07 | 67.12 | 57.61 | 31.83 | |
| PODNet-CNNNumber of tasks (N)=52022.07 | 66.43 | 58.9 | — | |
| MnemonicsNumber of tasks (N)=52022.07 | 64.54 | 56.85 | 13.85 | |
| LUCIRNumber of tasks (N)=52022.07 | 64.45 | 56.6 | 24.08 | |
| PODNet-CNNNumber of tasks (N)=102022.07 | 63.21 | 55.7 | — | |
| MnemonicsNumber of tasks (N)=102022.07 | 63.01 | 54.99 | 15.82 | |
| BiCNumber of tasks (N)=52022.07 | 62.65 | — | 25.06 | |
| LUCIRNumber of tasks (N)=102022.07 | 61.57 | 51.7 | 27.29 | |
| BiCNumber of tasks (N)=102022.07 | 58.72 | — | 28.34 | |
| iCaRLNumber of tasks (N)=52022.07 | 51.5 | 34.2 | 26.03 | |
| iCaRLNumber of tasks (N)=102022.07 | 46.89 | 38.91 | 33.76 | |
| LwFNumber of tasks (N)=52022.07 | 44.35 | 34.2 | 48.7 | |
| LwFNumber of tasks (N)=102022.07 | 38.9 | 30.1 | 47.94 |