Image Classification on CIFAR-10-LT (IF 50)
91.1Top-1 AccuracyGLMC + Loss + LR
Evaluation results
| Method | Links | |
|---|---|---|
| GLMC + Loss + LRreweighting=Inverse-view, LR Schedule=MiLeLR2026.05 | 91.1 | |
| GLMC + Lossreweighting=Inverse-view2026.05 | 90.8 | |
| GLMCVenue=CVPR232026.05 | 90.2 | |
| GLMC + SEL2026.05 | 88.6 | |
| ProCoBackbone=ResNet-32, Training Epochs=2002024.03 | 88.2 | |
| OursBackbone=ResNet-322026.03 | 88.2 | |
| ResComBackbone=ResNet-32, Imbalanced Factor=502022.03 | 88 | |
| FeatReconBackbone=ResNet-322026.03 | 87.8 | |
| FeatReconVenue=ICLR252026.05 | 87.8 | |
| GBGBackbone=ResNet-322026.03 | 87.7 | |
| RBLVenue=ICML232026.05 | 87.6 | |
| BCLBackbone=ResNet-32, Training Epochs=2002024.03 | 87.2 | |
| BCLBackbone=ResNet-322026.03 | 87.2 | |
| DiffuLT + BBNBackbone=ResNet-322026.03 | 87.2 | |
| Logit AdjustmentBackbone=ResNet-32, Training Epochs=2002024.03 | 87.1 | |
| DiffuLTBackbone=ResNet-322026.03 | 86.9 | |
| GCL + CRBackbone Net=ResNet-322023.03 | 86.8 | |
| SELBackbone=ResNet-322026.03 | 86.3 | |
| MISLASBackbone=ResNet-32, Imbalanced Factor=502022.03 | 85.7 | |
| MISLASBackbone Net=ResNet-322023.03 | 85.7 | |
| MiSLASVenue=CVPR212026.05 | 85.7 | |
| CE* + Lossreweighting=Inverse-view2026.05 | 85.7 | |
| GCLBackbone Net=ResNet-322023.03 | 85.5 | |
| Hybrid-PSCBackbone=ResNet-32, Imbalanced Factor=502022.03 | 85.4 | |
| Hybrid-SCBackbone=ResNet-32, Training Epochs=2002024.03 | 85.4 | |
| HCLVenue=CVPR212026.05 | 85.4 | |
| CMO + Lossreweighting=Inverse-view2026.05 | 85.4 | |
| CE* + Loss + LRreweighting=Inverse-view, LR Schedule=MiLeLR2026.05 | 85.3 | |
| CMO + Loss + LRreweighting=Inverse-view, LR Schedule=MiLeLR2026.05 | 85.3 | |
| ResLTBackbone=ResNet-32, Training Epochs=2002024.03 | 85.2 | |
| Balanced SoftmaxBackbone=ResNet-32, Imbalanced Factor=502022.03 | 84.9 | |
| De-c-TDE + CRBackbone Net=ResNet-322023.03 | 84.5 | |
| MetaSAug-LDAMBackbone=ResNet-32, Training Epochs=2002024.03 | 84.3 | |
| RIDE-3 expertsVenue=ICLR212026.05 | 84 | |
| FedYoYoVenue=ICCV252026.05 | 83.9 | |
| Causal NormBackbone=ResNet-32, Imbalanced Factor=502022.03 | 83.6 | |
| Casual ModelBackbone=ResNet-32, Training Epochs=2002024.03 | 83.6 | |
| De-c-TDEBackbone Net=ResNet-322023.03 | 83.6 | |
| ResLTBackbone=ResNet-32, Imbalanced Factor=502022.03 | 83.5 | |
| BBN + CRBackbone Net=ResNet-322023.03 | 83.5 | |
| TSCBackbone=ResNet-32, Training Epochs=2002024.03 | 82.9 | |
| TSCVenue=CVPR222026.05 | 82.9 | |
| RSGBackbone=ResNet-32, Imbalanced Factor=502022.03 | 82.8 | |
| INC-DRWVenue=AISTATS232026.05 | 82.7 | |
| ELF+LDAMBackbone=ResNet-32, Imbalanced Factor=502022.03 | 82.4 | |
| BBNBackbone=ResNet-32, Imbalanced Factor=502022.03 | 82.2 | |
| BBNBackbone=ResNet-322026.03 | 82.2 | |
| BBNVenue=CVPR202026.05 | 82.2 | |
| SSPBackbone=ResNet-32, Training Epochs=2002024.03 | 82.1 | |
| BBNBackbone Net=ResNet-322023.03 | 82.1 | |
| Focal-SAMVenue=ICML252026.05 | 82 | |
| ETF-DR + Loss + LRreweighting=Inverse-view, LR Schedule=MiLeLR2026.05 | 81.8 | |
| KCLVenue=ICLR212026.05 | 81.7 | |
| CAM-BSBackbone=ResNet-322026.03 | 81.4 | |
| ETF-DR + DisA2026.05 | 81.4 | |
| ETF-DR + Lossreweighting=Inverse-view2026.05 | 81.4 | |
| BBNBackbone=ResNet-32, Training Epochs=2002024.03 | 81.2 | |
| LDAMBackbone=ResNet-32, Imbalanced Factor=502022.03 | 81.1 | |
| CE*reproduced=true2026.05 | 81.1 | |
| LDAM-DRWBackbone=ResNet-32, Training Epochs=2002024.03 | 81 | |
| LDAM-DRWBackbone=ResNet-322026.03 | 81 | |
| ETF-DRVenue=NeurIPS222026.05 | 81 | |
| cRTBackbone=ResNet-322026.03 | 80.4 | |
| CMOVenue=CVPR22, reproduced=true2026.05 | 80 | |
| CB Loss + CRBackbone Net=ResNet-322023.03 | 79.8 | |
| SELVenue=ICCV252026.05 | 79.8 | |
| CB-FocalBackbone=ResNet-32, Training Epochs=2002024.03 | 79.3 | |
| CB LossBackbone Net=ResNet-322023.03 | 79.2 | |
| Focal Loss + CRBackbone Net=ResNet-322023.03 | 79.1 | |
| DisAVenue=ICML242026.05 | 78.6 | |
| Focal LossBackbone Net=ResNet-322023.03 | 76.7 | |
| Focal LossBackbone=ResNet-322026.03 | 76.7 | |
| Cross Entropy + CRBackbone Net=ResNet-322023.03 | 76.2 | |
| MLLM-LLaVA-FLType=Our framework2024.09 | 76.11 | |
| CLIP2FLType=SOTA2024.09 | 75.35 | |
| Cross EntropyBackbone Net=ResNet-322023.03 | 74.8 | |
| CEBackbone=ResNet-322026.03 | 74.8 | |
| CReFFType=Classifier-retraining2024.09 | 73.08 | |
| CCVRType=Heterogeneity-oriented FL methods2024.09 | 71.89 | |
| Ratio LossType=Imbalance-oriented FL methods2024.09 | 64.77 | |
| FedNovaType=Heterogeneity-oriented FL methods2024.09 | 63.91 | |
| FedProxType=Heterogeneity-oriented FL methods2024.09 | 60.89 | |
| FedBEType=Heterogeneity-oriented FL methods2024.09 | 59.55 | |
| FedAvgType=Heterogeneity-oriented FL methods2024.09 | 59.36 | |
| FedDFType=Heterogeneity-oriented FL methods2024.09 | 58.74 | |
| Fed-Focal LossType=Imbalance-oriented FL methods2024.09 | 57.42 | |
| FedAvgMType=Heterogeneity-oriented FL methods2024.09 | 57.11 | |
| FedAvg+ τ-normType=Imbalance-oriented FL methods2024.09 | 51.41 |