Image Classification on CIFAR10 Corrupted
81.35AccuracyFSVI Ensemble
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FSVI EnsembleBackbone=ResNet-18, Context Distribution (pxc)=random monochrome2023.12 | 81.35 | — | — | |
| Deep EnsembleBackbone=ResNet-182023.12 | 81.22 | — | — | |
| FSVIBackbone=ResNet-18, Context Distribution (pxc)=CIFAR-1002023.12 | 81.2 | — | — | |
| FSVIBackbone=ResNet-18, Context Distribution (pxc)=random monochrome2023.12 | 80.81 | — | — | |
| MC DROPOUTBackbone=ResNet-182023.12 | 80.13 | — | — | |
| MFVI (tempered)Backbone=ResNet-182023.12 | 79.86 | — | — | |
| MFVIBackbone=ResNet-182023.12 | 79.36 | — | — | |
| MAPBackbone=ResNet-182023.12 | 78.87 | — | — | |
| Deep EnsemblesPass Type=Multi-Pass, Backbone=ResNet-28-102021.10 | 77.9 | 0.087 | 0.815 | |
| GSD (Ensembled)Pass Type=Multi-Pass, Backbone=ResNet-28-10, Optimization Strategy=β' Grid-Searched2021.10 | 77.9 | 0.069 | 0.773 | |
| LNBackbone=WRN22-102025.01 | 76.43 | — | — | |
| SWAGBackbone=ResNet-182023.12 | 76.12 | — | — | |
| CEBackbone=WRN22-102025.01 | 76.1 | — | — | |
| MAXBackbone=ResNet-18, Training Protocol=training from scratch2024.02 | 75.1 | — | — | |
| HEM-Backbone=WRN22-102025.01 | 75.08 | — | — | |
| GSD (Grid-Searched)Pass Type=Single-Pass, Backbone=ResNet-28-10, Optimization Strategy=β' Grid-Searched2021.10 | 74.9 | 0.067 | 0.826 | |
| GSD (Optimized)Pass Type=Single-Pass, Backbone=ResNet-28-10, Optimization Strategy=β' Optimized2021.10 | 74.9 | 0.085 | 0.853 | |
| SNGPPass Type=Single-Pass, Backbone=ResNet-28-102021.10 | 74.6 | 0.09 | 0.935 | |
| MMBackbone=WRN22-102025.01 | 74.31 | — | — | |
| RAMPBackbone=ResNet-18, Training Protocol=training from scratch2024.02 | 74.3 | — | — | |
| E-ATBackbone=ResNet-18, Training Protocol=training from scratch2024.02 | 73.8 | — | — | |
| DICEBackbone=WRN22-102025.01 | 73.61 | — | — | |
| LNBackbone=ResNet322025.01 | 73.04 | — | — | |
| VanillaPass Type=Single-Pass, Backbone=ResNet-28-102021.10 | 72.9 | 0.153 | 1.059 | |
| CEBackbone=ResNet322025.01 | 72.88 | — | — | |
| HEM-Backbone=ResNet322025.01 | 72.82 | — | — | |
| DUQPass Type=Single-Pass, Backbone=ResNet-28-102021.10 | 71.6 | 0.183 | 1.348 | |
| MMBackbone=ResNet322025.01 | 70.66 | — | — | |
| MC DropoutPass Type=Multi-Pass, Backbone=ResNet-28-102021.10 | 70 | 0.116 | 1.152 | |
| DICEBackbone=ResNet322025.01 | 67.16 | — | — | |
| CEBackbone=MobileNetS2025.01 | 62.71 | — | — | |
| MMBackbone=MobileNetS2025.01 | 61.03 | — | — | |
| DICEBackbone=MobileNetS2025.01 | 59.02 | — | — | |
| HEM-Backbone=MobileNetS2025.01 | 58.02 | — | — | |
| LNBackbone=MobileNetS2025.01 | 48.46 | — | — |