Out-of-Distribution Detection on CIFAR-10 Interp.
0.86AUROCVERA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VERAScore Function=log p_theta(x), Training Dataset=CIFAR102022.09 | 0.86 | — | |
| SADA-JEMScore Function=log p_theta(x), Training Dataset=CIFAR10, K=202022.09 | 0.83 | — | |
| HDGETraining Protocol=Supervised2020.10 | 0.82 | — | |
| SADA-JEMScore Function=log p_theta(x), Training Dataset=CIFAR10, K=102022.09 | 0.81 | — | |
| SADA-JEMScore Function=max_y p_theta(y|x), Training Dataset=CIFAR10, K=202022.09 | 0.8 | — | |
| Projection RegretCategory=Diffusion model2023.12 | 0.8 | 1.5 | |
| CSINetwork=ResNet-182020.07 | 0.793 | — | |
| SADA-JEMScore Function=log p_theta(x), Training Dataset=CIFAR10, K=52022.09 | 0.79 | — | |
| SADA-JEMScore Function=max_y p_theta(y|x), Training Dataset=CIFAR10, K=102022.09 | 0.78 | — | |
| WideResNetScore Function=max_y p_theta(y|x), Training Dataset=CIFAR102022.09 | 0.77 | — | |
| JEM++Score Function=max_y p_theta(y|x), Training Dataset=CIFAR10, M=202022.09 | 0.77 | — | |
| SADA-JEMScore Function=max_y p_theta(y|x), Training Dataset=CIFAR10, K=52022.09 | 0.77 | — | |
| JEMScore Function=max_y p_theta(y|x), Training Dataset=CIFAR10, K=202022.09 | 0.75 | — | |
| CLELCategory=EBM2023.12 | 0.72 | 2 | |
| PixelCNN++Category=Autoregressive2023.12 | 0.71 | 4.25 | |
| LikelihoodNetwork=EBM2020.07 | 0.7 | — | |
| IGEBMTraining Protocol=Unsupervised2020.10 | 0.7 | — | |
| VAEBMTraining Protocol=Unsupervised2020.10 | 0.7 | — | |
| IGEBMScore Function=log p_theta(x), Training Dataset=CIFAR102022.09 | 0.7 | — | |
| IGEBMCategory=EBM2023.12 | 0.7 | 5.25 | |
| VAEBMCategory=VAE + EBM2023.12 | 0.7 | 4.33 | |
| IGEBMScore Function=max_y p_theta(y|x), Training Dataset=CIFAR102022.09 | 0.69 | — | |
| Rot+TransNetwork=ResNet-182020.07 | 0.681 | — | |
| JEMTraining Protocol=Supervised2020.10 | 0.65 | — | |
| JEMScore Function=log p_theta(x), Training Dataset=CIFAR10, K=202022.09 | 0.65 | — | |
| Improved CDCategory=EBM2023.12 | 0.65 | 3.25 | |
| RotNetwork=ResNet-182020.07 | 0.64 | — | |
| NVAETraining Protocol=Unsupervised2020.10 | 0.64 | — | |
| NVAECategory=VAE2023.12 | 0.64 | 6.33 | |
| GOADNetwork=ResNet-182020.07 | 0.594 | — | |
| LikelihoodNetwork=Glow2020.07 | 0.582 | — | |
| JEM++Score Function=log p_theta(x), Training Dataset=CIFAR10, M=202022.09 | 0.57 | — | |
| LikelihoodNetwork=PixelCNN++2020.07 | 0.526 | — | |
| GlowTraining Protocol=Unsupervised2020.10 | 0.51 | — | |
| GLOWCategory=Flow-based2023.12 | 0.51 | 7.25 |