Image Classification on MNIST permutation-invariant (test)
212Incorrect Test Examples CountVirtual Adversarial
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Virtual AdversarialNumber of labeled samples=1002016.06 | 212 | — | |
| Ensemble of 10 of our modelsNumber of labeled samples=2002016.06 | 814.3 | — | |
| Ensemble of 10 of our modelsNumber of labeled samples=1002016.06 | 865.6 | — | |
| Our modelNumber of labeled samples=2002016.06 | 904.2 | — | |
| Our modelNumber of labeled samples=1002016.06 | 936.5 | — | |
| Auxiliary Deep Generative ModelNumber of labeled samples=1002016.06 | 962 | — | |
| Skip Deep Generative ModelNumber of labeled samples=1002016.06 | 1,327 | — | |
| Ladder networkNumber of labeled samples=1002016.06 | 10,637 | — | |
| Ensemble of 10 of our modelsNumber of labeled samples=502016.06 | 14,296 | — | |
| CatGANNumber of labeled samples=1002016.06 | 19,110 | — | |
| DGNNumber of labeled samples=1002016.06 | 33,314 | — | |
| Our modelNumber of labeled samples=502016.06 | 221,136 | — | |
| Ensemble of 10 of our modelsNumber of labeled samples=202016.06 | 1,134,445 | — | |
| Our modelNumber of labeled samples=202016.06 | 1,677,452 | — | |
| Deep L2-SVM2015.11 | — | 0.87 | |
| Maxout Networks2015.11 | — | 0.94 |