Image Classification on CIFAR-10-LT IF 10
92Top-1 AccuracyResCom
Evaluation results
| Method | Links | |
|---|---|---|
| ResComBackbone=ResNet-32, Imbalanced Factor=102022.03 | 92 | |
| ProCoBackbone=ResNet-32, Training Epochs=2002024.03 | 91.9 | |
| Balanced SoftmaxBackbone=ResNet-32, Imbalanced Factor=102022.03 | 91.3 | |
| Hybrid-PSCBackbone=ResNet-32, Imbalanced Factor=102022.03 | 91.1 | |
| Hybrid-SCBackbone=ResNet-32, Training Epochs=2002024.03 | 91.1 | |
| BCLBackbone=ResNet-32, Training Epochs=2002024.03 | 91.1 | |
| Logit AdjustmentBackbone=ResNet-32, Training Epochs=2002024.03 | 90.9 | |
| MISLASBackbone=ResNet-32, Imbalanced Factor=102022.03 | 90 | |
| MISLASBackbone Net=ResNet-322023.03 | 90 | |
| De-c-TDE + CRBackbone Net=ResNet-322023.03 | 89.9 | |
| MetaSAug-LDAMBackbone=ResNet-32, Training Epochs=2002024.03 | 89.7 | |
| ResLTBackbone=ResNet-32, Training Epochs=2002024.03 | 89.7 | |
| Cross Entropy + CRBackbone Net=ResNet-322023.03 | 89.5 | |
| BBN + CRBackbone Net=ResNet-322023.03 | 89.4 | |
| ResLTBackbone=ResNet-32, Imbalanced Factor=102022.03 | 89.1 | |
| CB Loss + CRBackbone Net=ResNet-322023.03 | 89.1 | |
| TSCBackbone=ResNet-32, Training Epochs=2002024.03 | 88.7 | |
| Causal NormBackbone=ResNet-32, Imbalanced Factor=102022.03 | 88.5 | |
| SSPBackbone=ResNet-32, Training Epochs=2002024.03 | 88.5 | |
| Casual ModelBackbone=ResNet-32, Training Epochs=2002024.03 | 88.5 | |
| De-c-TDEBackbone Net=ResNet-322023.03 | 88.5 | |
| BBNBackbone=ResNet-32, Imbalanced Factor=102022.03 | 88.4 | |
| Focal Loss + CRBackbone Net=ResNet-322023.03 | 88.4 | |
| BBNBackbone=ResNet-32, Training Epochs=2002024.03 | 88.3 | |
| BBNBackbone Net=ResNet-322023.03 | 88.3 | |
| LDAMBackbone=ResNet-32, Imbalanced Factor=102022.03 | 88.2 | |
| LDAM-DRWBackbone=ResNet-32, Training Epochs=2002024.03 | 88.2 | |
| LDAM-DRWBackbone Net=ResNet-322023.03 | 88.2 | |
| ELF+LDAMBackbone=ResNet-32, Imbalanced Factor=102022.03 | 88 | |
| CB-FocalBackbone=ResNet-32, Training Epochs=2002024.03 | 87.5 | |
| CB LossBackbone Net=ResNet-322023.03 | 87.4 | |
| Focal LossBackbone Net=ResNet-322023.03 | 86.6 | |
| Cross EntropyBackbone Net=ResNet-322023.03 | 86.3 | |
| MLLM-LLaVA-FLType=Our framework2024.09 | 81.45 | |
| CLIP2FLType=SOTA2024.09 | 81.18 | |
| CReFFType=Classifier-retraining2024.09 | 80.71 | |
| CCVRType=Heterogeneity-oriented FL methods2024.09 | 78.48 | |
| Ratio LossType=Imbalance-oriented FL methods2024.09 | 78.14 | |
| FedNovaType=Heterogeneity-oriented FL methods2024.09 | 77.79 | |
| FedBEType=Heterogeneity-oriented FL methods2024.09 | 77.78 | |
| FedAvgType=Heterogeneity-oriented FL methods2024.09 | 77.45 | |
| FedProxType=Heterogeneity-oriented FL methods2024.09 | 76.53 | |
| FedDFType=Heterogeneity-oriented FL methods2024.09 | 76.51 | |
| Fed-Focal LossType=Imbalance-oriented FL methods2024.09 | 73.74 | |
| FedAvg+ τ-normType=Imbalance-oriented FL methods2024.09 | 72.08 | |
| FedAvgMType=Heterogeneity-oriented FL methods2024.09 | 70.81 |