Image Classification on CIFAR-10 clean standard (test)
93.55AccuracyUniform SGD
Evaluation results
| Method | Links | |
|---|---|---|
| Uniform SGDBackbone=VGG-19, Training Epochs=1602026.05 | 93.55 | |
| DR-IS-onlineBackbone=VGG-19, Training Epochs=1602026.05 | 92.4 | |
| Standard ISBackbone=VGG-19, Training Epochs=1602026.05 | 90.81 | |
| Three-way HybridArchitecture=MLP2026.03 | 55.21 | |
| F-Mean HybridArchitecture=MLP2026.03 | 55.17 | |
| Gaussian HybridArchitecture=MLP2026.03 | 54.3 | |
| BaselineArchitecture=MLP2026.03 | 52.3 | |
| RHO-LOSS (faithful)Backbone=VGG-19, Training Epochs=160, Proxy Backbone=ResNet-20, Holdout size=5k, Refresh Protocol=recomputed every 5 epochs2026.05 | 36.32 | |
| RHO-LOSS (approx.)Backbone=VGG-19, Training Epochs=160, Proxy type=mean loss, Refresh=no refresh2026.05 | 20.21 |