Image Classification on CIFAR-10 (test) (Accuracy, Calibration, and Agreement)
98.77AccuracyE2OAL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| E2OALMismatch ratio=20%2026.03 | 98.77 | — | — | |
| E2OALMismatch ratio=30%2026.03 | 97.52 | — | — | |
| E2OAL*Mismatch ratio=20%, leveraging labeled unknowns=false2026.03 | 97.33 | — | — | |
| EAOAMismatch ratio=20%2026.03 | 97.23 | — | — | |
| BUALMismatch ratio=20%2026.03 | 96.48 | — | — | |
| EOALMismatch ratio=20%2026.03 | 96.23 | — | — | |
| E2OAL*Mismatch ratio=30%, leveraging labeled unknowns=false2026.03 | 95.94 | — | — | |
| EAOAMismatch ratio=30%2026.03 | 95.88 | — | — | |
| SGDFBackbone=DenseNet1212026.03 | 95.74 | — | — | |
| UncertaintyMismatch ratio=20%2026.03 | 95.7 | — | — | |
| E2OALMismatch ratio=40%2026.03 | 95.69 | — | — | |
| AdaBeliefBackbone=DenseNet1212026.03 | 95.44 | — | — | |
| SGDFBackbone=ResNet342026.03 | 95.33 | — | — | |
| AdaBeliefBackbone=ResNet342026.03 | 95.18 | — | — | |
| MQNetMismatch ratio=20%2026.03 | 95.12 | — | — | |
| AdaBoundBackbone=DenseNet1212026.03 | 95.06 | — | — | |
| BUALMismatch ratio=30%2026.03 | 95.04 | — | — | |
| MSVAGBackbone=DenseNet1212026.03 | 95.01 | — | — | |
| BADGEMismatch ratio=20%2026.03 | 94.95 | — | — | |
| AdaBoundBackbone=ResNet342026.03 | 94.93 | — | — | |
| MSVAGBackbone=ResNet342026.03 | 94.79 | — | — | |
| SGDBackbone=ResNet342026.03 | 94.62 | — | — | |
| AdamWBackbone=DenseNet1212026.03 | 94.61 | — | — | |
| SophiaGBackbone=DenseNet1212026.03 | 94.53 | — | — | |
| SGDBackbone=DenseNet1212026.03 | 94.52 | — | — | |
| RandomMismatch ratio=20%2026.03 | 94.48 | — | — | |
| AdamWBackbone=ResNet342026.03 | 94.42 | — | — | |
| Ensemble2022.06 | 94.41 | 86.08 | 5.56 | |
| AdamBackbone=DenseNet1212026.03 | 94.37 | — | — | |
| AdamBackbone=ResNet342026.03 | 94.3 | — | — | |
| LionBackbone=DenseNet1212026.03 | 94.26 | — | — | |
| CoresetMismatch ratio=20%2026.03 | 94.2 | — | — | |
| SophiaGBackbone=ResNet342026.03 | 94.15 | — | — | |
| MSPMismatch ratio=20%2026.03 | 94.15 | — | — | |
| LfOSAMismatch ratio=20%2026.03 | 94.15 | — | — | |
| LionBackbone=ResNet342026.03 | 93.97 | — | — | |
| RAdamBackbone=DenseNet1212026.03 | 93.75 | — | — | |
| EOALMismatch ratio=30%2026.03 | 93.64 | — | — | |
| FED2022.06 | 93.61 | 93.34 | 0.66 | |
| RAdamBackbone=ResNet342026.03 | 93.41 | — | — | |
| E2OAL*Mismatch ratio=40%, leveraging labeled unknowns=false2026.03 | 93.13 | — | — | |
| EAOAMismatch ratio=40%2026.03 | 93.09 | — | — | |
| BUALMismatch ratio=40%2026.03 | 92.52 | — | — | |
| DP-FedAdamWBackbone=Swin-Base, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 91.79 | — | — | |
| EOALMismatch ratio=40%2026.03 | 91.63 | — | — | |
| SGDFBackbone=VGG112026.03 | 91.61 | — | — | |
| MSPMismatch ratio=30%2026.03 | 91.51 | — | — | |
| DP-FedSAMBackbone=Swin-Base, alpha=0.6, Privacy Budget=(2ε, δ)2026.02 | 91.25 | — | — | |
| AdaBeliefBackbone=VGG112026.03 | 91.24 | — | — | |
| RandomMismatch ratio=30%2026.03 | 91.11 | — | — | |
| Hydra2022.06 | 91.07 | 94.6 | 5.39 | |
| EnDDAuxiliary Data=Unsupervised data used2022.06 | 91.05 | 100 | 5.63 | |
| EnDDAuxiliary Data=None2022.06 | 91.01 | 99.99 | 6.39 | |
| DP-FedAvg-LSBackbone=Swin-Base, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 90.97 | — | — | |
| BADGEMismatch ratio=30%2026.03 | 90.91 | — | — | |
| LfOSAMismatch ratio=30%2026.03 | 90.91 | — | — | |
| DP-LocalAdamWBackbone=Swin-Base, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 90.84 | — | — | |
| DP-FedAdamWBackbone=Swin-Base, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 90.68 | — | — | |
| RAdamBackbone=VGG112026.03 | 90.47 | — | — | |
| AdaBoundBackbone=VGG112026.03 | 90.41 | — | — | |
| DM Gaussian2022.06 | 90.27 | 95.88 | 2.48 | |
| DP-FedAvg-LSBackbone=Swin-Base, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 90.14 | — | — | |
| MSVAGBackbone=VGG112026.03 | 90.08 | — | — | |
| DP-SCAFFOLDBackbone=Swin-Base, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 90.07 | — | — | |
| SGDBackbone=VGG112026.03 | 89.83 | — | — | |
| DP-FedSAMBackbone=Swin-Base, alpha=0.1, Privacy Budget=(2ε, δ)2026.02 | 89.82 | — | — | |
| UncertaintyMismatch ratio=30%2026.03 | 89.77 | — | — | |
| DP-FedAvgBackbone=Swin-Base, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 89.56 | — | — | |
| CoresetMismatch ratio=30%2026.03 | 89.56 | — | — | |
| DM ConfODS2022.06 | 89.55 | 95.49 | 2.59 | |
| MQNetMismatch ratio=30%2026.03 | 89.39 | — | — | |
| DM ODS2022.06 | 89.35 | 95.51 | 2.65 | |
| DP-SCAFFOLDBackbone=Swin-Base, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 89.08 | — | — | |
| DP-LocalAdamWBackbone=Swin-Base, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 89.08 | — | — | |
| AdamWBackbone=VGG112026.03 | 88.59 | — | — | |
| SophiaGBackbone=VGG112026.03 | 88.53 | — | — | |
| DP-FedAvgBackbone=Swin-Base, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 88.23 | — | — | |
| AdamBackbone=VGG112026.03 | 88.12 | — | — | |
| LionBackbone=VGG112026.03 | 88.04 | — | — | |
| LfOSAMismatch ratio=40%2026.03 | 87.43 | — | — | |
| MQNetMismatch ratio=40%2026.03 | 87.42 | — | — | |
| MSPMismatch ratio=40%2026.03 | 87.21 | — | — | |
| RandomMismatch ratio=40%2026.03 | 87.18 | — | — | |
| BADGEMismatch ratio=40%2026.03 | 87.12 | — | — | |
| CoresetMismatch ratio=40%2026.03 | 86.38 | — | — | |
| DP-FedAdamWBackbone=Swin-Tiny, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 84.23 | — | — | |
| DP-FedSAMBackbone=Swin-Tiny, alpha=0.6, Privacy Budget=(2ε, δ)2026.02 | 84.05 | — | — | |
| DP-FedAvg-LSBackbone=Swin-Tiny, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 83.94 | — | — | |
| AMTM MD2022.06 | 83.81 | 12.95 | 64 | |
| DP-LocalAdamWBackbone=Swin-Tiny, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 83.73 | — | — | |
| UncertaintyMismatch ratio=40%2026.03 | 83.61 | — | — | |
| DP-FedAvgBackbone=Swin-Tiny, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 82.93 | — | — | |
| DP-SCAFFOLDBackbone=Swin-Tiny, alpha=0.6, Privacy Budget=(ε, δ)2026.02 | 82.85 | — | — | |
| DP-FedAdamWBackbone=Swin-Tiny, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 81.23 | — | — | |
| DP-FedSAMBackbone=Swin-Tiny, alpha=0.1, Privacy Budget=(2ε, δ)2026.02 | 80.36 | — | — | |
| DP-FedAvg-LSBackbone=Swin-Tiny, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 79.67 | — | — | |
| DP-LocalAdamWBackbone=Swin-Tiny, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 79.28 | — | — | |
| DP-SCAFFOLDBackbone=Swin-Tiny, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 78.53 | — | — | |
| DP-FedAvgBackbone=Swin-Tiny, alpha=0.1, Privacy Budget=(ε, δ)2026.02 | 78.24 | — | — | |
| FRePoTrain Arch=Conv-BN, Evaluation Architecture=Conv, Images per class=102022.06 | 65.5 | — | — |