Image Classification on CIFAR-100 400 labels
31.39Error RateSemi-SST
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Semi-SSTBackbone=ViT-Small, Pre-training=DINO2025.05 | 31.39 | — | — | |
| JEPAMatchIter.=2^172026.04 | 34.25 | — | — | |
| RegMixMatch*Iter.=2^202026.04 | 35.27 | — | — | |
| Suvio&Daino*Iter.=2^202026.04 | 35.4 | — | — | |
| Super-SSTBackbone=ViT-Small, Pre-training=DINO2025.05 | 35.5 | — | — | |
| SimMatch2025.05 | 37.08 | — | — | |
| SoftMatchIter.=2^202026.04 | 37.1 | — | — | |
| FlexMatch2025.05 | 37.76 | — | — | |
| SimMatchIter.=2^202026.04 | 37.81 | — | — | |
| SequenceMatchBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 37.86 | — | — | |
| FreeMatchBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 37.98 | — | — | |
| FreeMatchIter.=2^202026.04 | 37.98 | — | — | |
| FlatMatchIter.=2^202026.04 | 38.76 | — | — | |
| SimMatchV22025.05 | 39.32 | — | — | |
| CrMatchIter.=2^202026.04 | 39.45 | — | — | |
| FlexMatch2021.10 | 39.94 | — | — | |
| FlexMatchBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 39.94 | — | — | |
| FlexMatchIter.=2^202026.04 | 39.94 | — | — | |
| AdaMatch2025.05 | 40.73 | — | — | |
| UDA2025.05 | 41.6 | — | — | |
| DoubleMatchBackbone=Wide ResNet-28-8, Training Labels=400, Evaluation Protocol=min2022.05 | 41.83 | — | — | |
| DoubleMatchBackbone=Wide ResNet-28-8, Training Labels=400, Evaluation Protocol=last 202022.05 | 42.61 | — | — | |
| Self-Tuning+HCRBackbone=EfficientNet-B22022.06 | 42.71 | — | — | |
| ReMixMatch2021.10 | 42.75 | — | — | |
| ReMixMatchBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 42.75 | — | — | |
| ReMixMatchIter.=2^202026.04 | 42.75 | — | — | |
| DP-SSLBackbone=Wide ResNet-28-8, Training Labels=4002022.05 | 43.17 | — | — | |
| ReMixMatchBackbone=Wide ResNet-28-8, Training Labels=4002022.05 | 44.28 | — | — | |
| DASHBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 44.82 | — | — | |
| Dash (CTA)Backbone=Wide ResNet-28-8, Training Labels=4002022.05 | 44.83 | — | — | |
| Dash2025.05 | 45.38 | — | — | |
| FixMatch2025.05 | 45.48 | — | — | |
| MPLBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 46.26 | — | — | |
| UDA2021.10 | 46.39 | — | — | |
| UDABackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 46.39 | — | — | |
| UDAIter.=2^202026.04 | 46.39 | — | — | |
| FixMatch2021.10 | 46.42 | — | — | |
| FixMatchBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 46.42 | — | — | |
| FixMatchIter.=2^202026.04 | 46.42 | — | — | |
| Self-TuningBackbone=EfficientNet-B22022.06 | 47.17 | — | — | |
| CoMatch2025.05 | 47.49 | — | — | |
| FixMatch + CMW-Net2022.02 | 47.7 | — | — | |
| FixMatch2022.02 | 48.85 | — | — | |
| FixMatch (CTA)Backbone=Wide ResNet-28-8, Training Labels=4002022.05 | 49.95 | — | — | |
| CRMatch2025.05 | 53.41 | — | — | |
| CT+PLBackbone=EfficientNet-B22022.06 | 56.21 | — | — | |
| CT+MTBackbone=EfficientNet-B22022.06 | 56.78 | — | — | |
| Co-TuningBackbone=EfficientNet-B22022.06 | 57.58 | — | — | |
| FixMatchBackbone=EfficientNet-B22022.06 | 57.87 | — | — | |
| CT+FMBackbone=EfficientNet-B22022.06 | 57.94 | — | — | |
| DELTABackbone=EfficientNet-B22022.06 | 58.23 | — | — | |
| UDABackbone=EfficientNet-B22022.06 | 58.32 | — | — | |
| BSSBackbone=EfficientNet-B22022.06 | 58.49 | — | — | |
| L2-SPBackbone=EfficientNet-B22022.06 | 59.21 | — | — | |
| Pseudo LabelingBackbone=EfficientNet-B22022.06 | 59.21 | — | — | |
| UDABackbone=Wide ResNet-28-8, Training Labels=4002022.05 | 59.28 | — | — | |
| UDA2022.02 | 59.28 | — | — | |
| SimCLRv2Backbone=EfficientNet-B22022.06 | 59.45 | — | — | |
| Π-modelBackbone=EfficientNet-B22022.06 | 60.5 | — | — | |
| Mean TeacherBackbone=EfficientNet-B22022.06 | 60.68 | — | — | |
| Fine-TuningBackbone=EfficientNet-B22022.06 | 60.79 | — | — | |
| ReMixMatch2025.05 | 64.91 | — | — | |
| MixMatch2021.10 | 67.59 | — | — | |
| MixMatchBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 67.59 | — | — | |
| MixMatchIter.=2^202026.04 | 67.59 | — | — | |
| MixMatchBackbone=Wide ResNet-28-8, Training Labels=4002022.05 | 67.61 | — | — | |
| MixMatch2022.02 | 67.61 | — | — | |
| VAT2025.05 | 79.96 | — | — | |
| Mean Teacher2021.10 | 81.11 | — | — | |
| MeanTeacherBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 81.11 | — | — | |
| MeanTeacherIter.=2^202026.04 | 81.11 | — | — | |
| MixMatch2025.05 | 83.6 | — | — | |
| VAT2021.10 | 85.2 | — | — | |
| VATBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 85.2 | — | — | |
| Π-Model2021.10 | 86.96 | — | — | |
| Pi ModelBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 86.96 | — | — | |
| Pseudo-Labeling2021.10 | 87.45 | — | — | |
| Pseudo LabelBackbone=Wide-ResNet-28-8, #LABEL=4002023.10 | 87.45 | — | — | |
| PseudoLabelIter.=2^202026.04 | 87.45 | — | — | |
| Pseudo-Labeling2025.05 | 87.67 | — | — | |
| Mean Teacher2025.05 | 88.18 | — | — | |
| Pi-Model2025.05 | 89.6 | — | — | |
| FlexMatchBackbone=ViT [Dosovitskiy et al., 2021] / ResNet-50 [He et al., 2016], SemiReward=false2023.10 | — | 17.8 | — | |
| FlexMatch + SemiRewardBackbone=ViT [Dosovitskiy et al., 2021] / ResNet-50 [He et al., 2016], SemiReward=true2023.10 | — | 17.59 | 0.63 | |
| Pseudo LabelBackbone=ViT [Dosovitskiy et al., 2021] / ResNet-50 [He et al., 2016], SemiReward=false2023.10 | — | 25.16 | — | |
| Pseudo Label + SemiRewardBackbone=ViT [Dosovitskiy et al., 2021] / ResNet-50 [He et al., 2016], SemiReward=true2023.10 | — | 23.84 | 0.63 | |
| SoftMatch/FreeMatchBackbone=ViT [Dosovitskiy et al., 2021] / ResNet-50 [He et al., 2016], SemiReward=false2023.10 | — | 15.97 | — | |
| SoftMatch/FreeMatch + SemiRewardBackbone=ViT [Dosovitskiy et al., 2021] / ResNet-50 [He et al., 2016], SemiReward=true2023.10 | — | 15.62 | 0.63 |