Semi-supervised Image Classification on CIFAR-10 4000 labeled samples (train test)
13.1Test ErrorS4L + ForkMerge+
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| S4L + ForkMerge+baseline=S4L, ATL approach=ForkMerge+2023.01 | 13.1 | — | 46.3 | |
| S4L + ARMLbaseline=S4L, ATL approach=ARML2023.01 | 13.7 | — | 43.2 | |
| S4L + Grid Searchbaseline=S4L, ATL approach=Grid Search2023.01 | 13.8 | — | 42.3 | |
| VAT2023.01 | 13.9 | — | 43.8 | |
| S4L + GradNormbaseline=S4L, ATL approach=GradNorm2023.01 | 14.1 | — | 35.3 | |
| S4L + Auto-λbaseline=S4L, ATL approach=Auto-λ2023.01 | 14.2 | — | 41 | |
| S4L + GCSbaseline=S4L, ATL approach=GCS2023.01 | 15 | — | 35.6 | |
| S4L + EWbaseline=S4L, ATL approach=EW2023.01 | 15.7 | — | 30.7 | |
| S4L + Post-trainbaseline=S4L, ATL approach=Post-train2023.01 | 15.8 | — | 30.4 | |
| Mean Teacher2023.01 | 15.9 | — | 38.8 | |
| S4L + OL_AUXbaseline=S4L, ATL approach=OL_AUX2023.01 | 16.1 | — | 29.8 | |
| Π-Model2023.01 | 16.4 | — | 31.5 | |
| Pseudo-Label2023.01 | 17.8 | — | 26.4 | |
| STL2023.01 | 20.3 | — | — | |
| FixMatchBackbone=WRN-28-22021.01 | — | 95.7 | — | |
| ICTBackbone=WRN-28-22021.01 | — | 92.7 | — | |
| LGA+VATBackbone=WRN-28-22021.01 | — | 87.9 | — | |
| MixMatchBackbone=WRN-28-22021.01 | — | 93.8 | — | |
| ReMixMatchBackbone=WRN-28-22021.01 | — | 94.9 | — | |
| UDABackbone=WRN-28-22021.01 | — | 94.9 | — | |
| UDA+AutoDropoutBackbone=WRN-28-22021.01 | — | 95.8 | — |