Classification on CIFAR-100 (test)
84.52AccuracyF-Deep-ens
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| F-Deep-ensBackbone=WideResNet28x10, Category=Variational inference2023.02 | 84.52 | 0.5644 | 1.91 | |
| F-SWAGBackbone=WideResNet28x10, Category=Sample2023.02 | 83.57 | 0.5757 | 1.96 | |
| F-SWAG-DiagBackbone=WideResNet28x10, Category=Sample2023.02 | 83.5 | 0.5763 | 1.51 | |
| F-MC-DropoutBackbone=WideResNet28x10, Category=Variational inference2023.02 | 83.24 | 0.6144 | 2.5 | |
| Deep-ensBackbone=WideResNet28x10, Category=Variational inference2023.02 | 83.04 | 0.6958 | 4.83 | |
| F-Deep-ensBackbone=PreResNet-164, Category=Variational inference2023.02 | 82.54 | 0.6286 | 1.43 | |
| SWAG-DiagBackbone=WideResNet28x10, Category=Sample2023.02 | 82.4 | 0.615 | 3.22 | |
| MC-DropoutBackbone=WideResNet28x10, Category=Variational inference2023.02 | 82.3 | 0.65 | 5.74 | |
| SWAGBackbone=WideResNet28x10, Category=Sample2023.02 | 82.23 | 0.6078 | 1.13 | |
| F-SGLDBackbone=WideResNet28x10, Category=Markov chain Monte Carlo2023.02 | 82.12 | 0.6722 | 8.2 | |
| Deep-ensBackbone=PreResNet-164, Category=Variational inference2023.02 | 82.08 | 0.7189 | 3.34 | |
| Deep EnsembleBackbone=ResNet-34, Ensemble size (M)=42022.05 | 82 | — | 4.4 | |
| Deep EnsembleBackbone=ResNet-18, Ensemble size (M)=42022.05 | 81.6 | — | 4.1 | |
| SGLDBackbone=WideResNet28x10, Category=Markov chain Monte Carlo2023.02 | 81.38 | 0.7123 | 9.58 | |
| F-MC-DropoutBackbone=PreResNet-164, Category=Variational inference2023.02 | 81.06 | 0.7027 | 5.14 | |
| F-SWAG-DiagBackbone=PreResNet-164, Category=Sample2023.02 | 81.01 | 0.6645 | 2.42 | |
| F-SWAGBackbone=PreResNet-164, Category=Sample2023.02 | 80.93 | 0.6704 | 3.5 | |
| F-SGLDBackbone=PreResNet-164, Category=Markov chain Monte Carlo2023.02 | 80.82 | 0.7276 | 10.85 | |
| FiLM-EnsembleBackbone=ResNet-34, Ensemble size (M)=42022.05 | 80.2 | — | 4.5 | |
| SWAG-DiagBackbone=PreResNet-164, Category=Sample2023.02 | 80.18 | 0.6837 | 2.39 | |
| SageMixBackbone=PreActResNet18, Time (sec)=24.32022.10 | 80.16 | — | — | |
| SGLDBackbone=PreResNet-164, Category=Markov chain Monte Carlo2023.02 | 80.13 | 0.7604 | 11.61 | |
| Co-MixupBackbone=PreActResNet18, Time (sec)=147.02022.10 | 80.13 | — | — | |
| SWAGBackbone=PreResNet-164, Category=Sample2023.02 | 79.9 | 0.6595 | 5.87 | |
| MC-DropoutBackbone=PreResNet-164, Category=Variational inference2023.02 | 79.5 | 0.9162 | 9.93 | |
| FiLM-EnsembleBackbone=ResNet-18, Ensemble size (M)=42022.05 | 79.4 | — | 3.8 | |
| Puzzle MixBackbone=PreActResNet18, Time (sec)=34.92022.10 | 79.38 | — | — | |
| Single NetworkBackbone=ResNet-34, Ensemble size (M)=42022.05 | 79.3 | — | 8.9 | |
| SaliencyMixBackbone=PreActResNet18, Time (sec)=21.12022.10 | 79.06 | — | — | |
| CutMixBackbone=PreActResNet18, Time (sec)=23.42022.10 | 78.71 | — | — | |
| Manifold MixupBackbone=PreActResNet18, Time (sec)=20.82022.10 | 78.36 | — | — | |
| BatchEnsembleBackbone=ResNet-34, Ensemble size (M)=42022.05 | 78.3 | — | 5.6 | |
| Single NetworkBackbone=ResNet-18, Ensemble size (M)=42022.05 | 78 | — | 4.6 | |
| BatchEnsembleBackbone=ResNet-18, Ensemble size (M)=42022.05 | 77.7 | — | 5.2 | |
| MixupBackbone=PreActResNet18, Time (sec)=20.42022.10 | 77.57 | — | — | |
| VanillaBackbone=PreActResNet18, Time (sec)=13.12022.10 | 76.41 | — | — | |
| MC-DropoutBackbone=ResNet-18, Ensemble size (M)=42022.05 | 75.5 | — | 6.4 | |
| DnCBackbone=ResNet-50, Pre-training dataset=JFT, Pre-training epochs=4500 ImageNet-equivalent, Evaluation protocol=Linear transfer2022.01 | 74.9 | — | — | |
| ReLICv2Backbone=ResNet-50, Pre-training dataset=JFT, Pre-training epochs=5000 ImageNet-equivalent, Evaluation protocol=Linear transfer2022.01 | 73 | — | — | |
| ReLICv2Backbone=ResNet-50, Pre-training dataset=JFT, Pre-training epochs=1000 ImageNet-equivalent, Evaluation protocol=Linear transfer2022.01 | 72.6 | — | — | |
| MasksembleBackbone=ResNet-18, Ensemble size (M)=42022.05 | 72.5 | — | 7.5 | |
| BYOLBackbone=ResNet-50, Pre-training dataset=JFT, Pre-training epochs=5000 ImageNet-equivalent, Evaluation protocol=Linear transfer2022.01 | 72.4 | — | — | |
| MC-DropoutBackbone=ResNet-34, Ensemble size (M)=42022.05 | 72.2 | — | 7.9 | |
| AugerinoBase Group=SE(2), Number of Elements=8, Partial Equivariance=false, Augerino=true2021.10 | 70.8 | — | — | |
| MasksembleBackbone=ResNet-34, Ensemble size (M)=42022.05 | 70.1 | — | 6.7 | |
| AugerinoBase Group=SE(2), Number of Elements=4, Partial Equivariance=false, Augerino=true2021.10 | 69.83 | — | — | |
| Partial G-CNNBase Group=SE(2), Number of Elements=8, Partial Equivariance=true, Augerino=false2021.10 | 69.66 | — | — | |
| F-EDL2025.10 | 69.4 | — | — | |
| Partial G-CNNBase Group=SE(2), Number of Elements=4, Partial Equivariance=true, Augerino=false2021.10 | 68.99 | — | — | |
| Highway NetworkData Augmentation=typical2015.07 | 67.76 | — | — | |
| G-CNNBase Group=SE(2), Number of Elements=8, Partial Equivariance=false, Augerino=false2021.10 | 67.7 | — | — | |
| CNNBase Group=T(2), Number of Elements=1, Architecture=13-layer CNN2021.10 | 67.14 | — | — | |
| AugerinoBase Group=SE(2), Number of Elements=2, Partial Equivariance=false, Augerino=true2021.10 | 66.72 | — | — | |
| I-EDL2025.10 | 66.38 | — | — | |
| All-CNNData Augmentation=typical2015.07 | 66.29 | — | — | |
| dasNetData Augmentation=typical2015.07 | 66.22 | — | — | |
| DAEDL2025.10 | 66.01 | — | — | |
| G-CNNBase Group=SE(2), Number of Elements=4, Partial Equivariance=false, Augerino=false2021.10 | 65.97 | — | — | |
| Dropout2025.10 | 65.94 | — | — | |
| Partial G-CNNBase Group=SE(2), Number of Elements=2, Partial Equivariance=true, Augerino=false2021.10 | 65.79 | — | — | |
| DSNData Augmentation=typical2015.07 | 65.43 | — | — | |
| NiNData Augmentation=typical2015.07 | 64.32 | — | — | |
| R-EDL2025.10 | 63.53 | — | — | |
| G-CNNBase Group=SE(2), Number of Elements=2, Partial Equivariance=false, Augerino=false2021.10 | 62.06 | — | — | |
| MaxoutData Augmentation=typical2015.07 | 61.42 | — | — | |
| V-MCR2Objective=V-MCR2, Training Epochs=2000, d=5002022.03 | 58.72 | — | — | |
| CEObjective=CE, Training Epochs=2000, d=5002022.03 | 58.4 | — | — | |
| MIMOBackbone=ResNet-34, Ensemble size (M)=42022.05 | 56.2 | — | 13.2 | |
| WholeIPC=Full, Ratio=100%2025.12 | 56.2 | — | — | |
| GeoDMIPC=50, Ratio=10%2025.12 | 55 | — | — | |
| NCFMIPC=50, Ratio=10%2025.12 | 54.7 | — | — | |
| DSDMIPC=50, Ratio=10%2025.12 | 53.1 | — | — | |
| IIDIPC=50, Ratio=10%2025.12 | 51.3 | — | — | |
| ATTIPC=50, Ratio=10%2025.12 | 51.2 | — | — | |
| M3DIPC=50, Ratio=10%2025.12 | 50.9 | — | — | |
| FTDIPC=50, Ratio=10%2025.12 | 50.7 | — | — | |
| IDMIPC=50, Ratio=10%2025.12 | 50 | — | — | |
| GeoDMIPC=10, Ratio=2%2025.12 | 49.2 | — | — | |
| DREAMIPC=50, Ratio=10%2025.12 | 48.5 | — | — | |
| NCFMIPC=10, Ratio=2%2025.12 | 48.4 | — | — | |
| MIMOBackbone=ResNet-18, Ensemble size (M)=42022.05 | 48 | — | 8.3 | |
| MTTIPC=50, Ratio=10%2025.12 | 47.7 | — | — | |
| DSDMIPC=10, Ratio=2%2025.12 | 46 | — | — | |
| EDL2025.10 | 45.91 | — | — | |
| IIDIPC=10, Ratio=2%2025.12 | 45.7 | — | — | |
| G-VBSMIPC=50, Ratio=10%2025.12 | 45.7 | — | — | |
| IDMIPC=10, Ratio=2%2025.12 | 45.1 | — | — | |
| ATTIPC=10, Ratio=2%2025.12 | 44.2 | — | — | |
| DMIPC=50, Ratio=10%2025.12 | 43.6 | — | — | |
| FTDIPC=10, Ratio=2%2025.12 | 43.4 | — | — | |
| DREAMIPC=10, Ratio=2%2025.12 | 43 | — | — | |
| DSAIPC=50, Ratio=10%2025.12 | 42.8 | — | — | |
| M3DIPC=10, Ratio=2%2025.12 | 42.4 | — | — | |
| MTTIPC=10, Ratio=2%2025.12 | 40.1 | — | — | |
| DCCIPC=50, Ratio=10%2025.12 | 40 | — | — | |
| G-VBSMIPC=10, Ratio=2%2025.12 | 38.7 | — | — | |
| GeoDMIPC=1, Ratio=0.2%2025.12 | 38 | — | — | |
| CAFEIPC=50, Ratio=10%2025.12 | 37.9 | — | — | |
| NCFMIPC=1, Ratio=0.2%2025.12 | 34.4 | — | — | |
| HerdingIPC=50, Ratio=10%2025.12 | 33.7 | — | — |