Image Classification on MNIST Aff(2, R) 2D Affine (test)
96.95AccuracyResNet18 + TIED
Evaluation results
| Method | Links | |
|---|---|---|
| ResNet18 + TIEDEnergy=VAE evidence lower bound (+ adv. reg.)2026.02 | 96.95 | |
| affConv / homConv2026.02 | 95.08 | |
| ResNet18 + LieLACEnergy=VAE evidence lower bound (+ adv. reg.)2026.02 | 94.38 | |
| ResNet18 + FoCalEnergy=VAE evidence lower bound (+ adv. reg.)2026.02 | 86.38 | |
| ResNet18 + TIEDEnergy=Classifier logit confidence2026.02 | 82.64 | |
| ResNet18 + Kinetic LangevinEnergy=VAE evidence lower bound (+ adv. reg.)2026.02 | 74.72 | |
| ResNet18 + LieLACEnergy=Classifier logit confidence2026.02 | 73.79 | |
| ResNet18 + ITSEnergy=Classifier logit confidence2026.02 | 67.18 | |
| ResNet18 + FoCalEnergy=Classifier logit confidence2026.02 | 66.97 | |
| ResNet182026.02 | 55.48 | |
| ResNet18 + Kinetic LangevinEnergy=Classifier logit confidence2026.02 | 53.98 | |
| ResNet18 + ITSEnergy=VAE evidence lower bound (+ adv. reg.)2026.02 | 45.97 |