Class-Incremental Learning on Dissimilar Tasks (CIFAR-10, SVHN, STL-10, MNIST, Fashion-MNIST)
67Avg Acc (Af)SSIL
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SSILNormalization=TBBN, Backbone=ResNet-182022.01 | 67 | 71.77 | 9.74 | 72.86 | |
| SSILNormalization=BN, Backbone=ResNet-182022.01 | 66.15 | 70.62 | 10.13 | 72.59 | |
| SSILNormalization=CN, Backbone=ResNet-182022.01 | 65.64 | 70.56 | 9.16 | 69.66 | |
| EEILNormalization=TBBN, Backbone=ResNet-182022.01 | 64.88 | 72.8 | 26.67 | 91.55 | |
| LUCIRNormalization=TBBN, Backbone=ResNet-182022.01 | 63.64 | 73.72 | 26.73 | 90.23 | |
| FTNormalization=TBBN, Backbone=ResNet-182022.01 | 63.42 | 71.32 | 28.05 | 91.47 | |
| LUCIRNormalization=CN, Backbone=ResNet-182022.01 | 62.99 | 72.81 | 26.7 | 89.49 | |
| LUCIRNormalization=BN, Backbone=ResNet-182022.01 | 62.42 | 72.91 | 28.09 | 90.35 | |
| EEILNormalization=BN, Backbone=ResNet-182022.01 | 62.04 | 72.28 | 29.46 | 91.5 | |
| EEILNormalization=CN, Backbone=ResNet-182022.01 | 61.73 | 72.4 | 29.27 | 91 | |
| FTNormalization=CN, Backbone=ResNet-182022.01 | 61.41 | 71 | 29.9 | 91.4 | |
| FTNormalization=BN, Backbone=ResNet-182022.01 | 59.6 | 65.85 | 31.48 | 91.08 |