Class-Incremental Learning on ImageNet-100 10-task (test)
39.4Avg ForgettingFT
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| FTNormalization=BN, Backbone=ResNet-182022.01 | 39.4 | 59.6 | 48.02 | 87.42 | |
| EEILNormalization=BN, Backbone=ResNet-182022.01 | 40.06 | 61.15 | 47.78 | 87.84 | |
| LUCIRNormalization=CN, Backbone=ResNet-182022.01 | 40.44 | 61.44 | 42.04 | 83.48 | |
| FTNormalization=CN, Backbone=ResNet-182022.01 | 41.96 | 60.02 | 45.32 | 87.28 | |
| LUCIRNormalization=BN, Backbone=ResNet-182022.01 | 42.26 | 63.82 | 41.68 | 83.94 | |
| EEILNormalization=CN, Backbone=ResNet-182022.01 | 42.48 | 61.43 | 44.44 | 86.92 | |
| FTNormalization=TBBN, Backbone=ResNet-182022.01 | 43.2 | 61.69 | 43.62 | 87.82 | |
| LUCIRNormalization=TBBN, Backbone=ResNet-182022.01 | 43.72 | 64.36 | 40.18 | 83.9 | |
| EEILNormalization=TBBN, Backbone=ResNet-182022.01 | 45.18 | 63.48 | 42.66 | 87.84 | |
| PODNetNormalization=CN, Backbone=ResNet-182022.01 | 46.2 | 62.91 | 23.66 | 68.5 | |
| AFCNormalization=CN, Backbone=ResNet-182022.01 | 48 | 65.21 | 20.1 | 70.68 | |
| PODNetNormalization=BN, Backbone=ResNet-182022.01 | 49.05 | 65.41 | 22.4 | 69.99 | |
| PODNetNormalization=TBBN, Backbone=ResNet-182022.01 | 49.3 | 65.7 | 21.85 | 69.76 | |
| SSILNormalization=BN, Backbone=ResNet-182022.01 | 49.56 | 65.79 | 21.2 | 69.94 | |
| SSILNormalization=CN, Backbone=ResNet-182022.01 | 50.58 | 64.81 | 18.56 | 65.04 | |
| SSILNormalization=TBBN, Backbone=ResNet-182022.01 | 51.3 | 66.51 | 19.58 | 70.64 | |
| AFCNormalization=BN, Backbone=ResNet-182022.01 | 52.5 | 67.53 | 19.7 | 72.22 | |
| AFCNormalization=TBBN, Backbone=ResNet-182022.01 | 54 | 68.68 | 19 | 73.22 |