Class Incremental Learning on Tiny-ImageNet-200 T200-B100-S20 (test)
34.85Final AccuracyGIS (Ours)
Evaluation results
| Method | Links | |
|---|---|---|
| GIS (Ours)Backbone=ResNet-18, Batch size=64, Optimizer=Adam, Initial learning rate=0.001, Epochs per step=30, Memory buffer size=20 instances per class2023.04 | 34.85 | |
| HFABackbone=ResNet-18, Batch size=64, Optimizer=Adam, Initial learning rate=0.001, Epochs per step=30, Memory buffer size=20 instances per class2023.04 | 31.37 | |
| IL2ABackbone=ResNet-18, Batch size=64, Optimizer=Adam, Initial learning rate=0.001, Epochs per step=30, Memory buffer size=20 instances per class2023.04 | 28.34 | |
| EEILBackbone=ResNet-18, Batch size=64, Optimizer=Adam, Initial learning rate=0.001, Epochs per step=30, Memory buffer size=20 instances per class2023.04 | 27.64 | |
| iCaRLNMEBackbone=ResNet-18, Batch size=64, Optimizer=Adam, Initial learning rate=0.001, Epochs per step=30, Memory buffer size=20 instances per class2023.04 | 27.51 | |
| iCaRLCNNBackbone=ResNet-18, Batch size=64, Optimizer=Adam, Initial learning rate=0.001, Epochs per step=30, Memory buffer size=20 instances per class2023.04 | 20.28 |