Image Classification on CIFAR-100 (test) (k-Accuracy Profile)
64.8Accuracy (k=80)Fully Supervised
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Fully SupervisedLabel Usage=With Labels2026.05 | 64.8 | — | 77.9 | — | 74.7 | 69.3 | 54.5 | — | |
| Semi-AUM+CutoffLabel Usage=With 10% Labels2026.05 | 63.1 | — | 77.4 | — | 73.5 | 67.7 | 53.9 | — | |
| Semi-DUAL+BetaLabel Usage=With 10% Labels2026.05 | 62.9 | — | 77.1 | — | 73.8 | 67.3 | 53.1 | — | |
| ELFS (DINO)Label Usage=Without Labels2026.05 | 61.9 | — | 77.3 | — | 73.8 | 67.4 | 53 | — | |
| ZCoreLabel Usage=Without Labels2026.05 | 61.9 | — | 76 | — | 72.9 | 65.9 | 52.1 | — | |
| ELFS (Self-Encoder)Label Usage=Without Labels2026.05 | 59.5 | — | 75.5 | — | 71.7 | 64.8 | 46.9 | — | |
| RandomLabel Usage=Without Labels2026.05 | 59.2 | — | 75.2 | — | 71.7 | 64.9 | 45.1 | — | |
| Score ExtrapolationLabel Usage=With 10% Labels2026.05 | 53.8 | — | 75.5 | — | 70.8 | 61.9 | 34.9 | — | |
| Jo-SNC-SEMPublication=Ours2025.08 | 44.26 | 66.02 | — | 62.43 | — | — | — | — | |
| Jo-SNCPublication=IEEE TPAMI 20252025.08 | 43.45 | 65.49 | — | 61.59 | — | — | — | — | |
| PrototypicalityLabel Usage=Without Labels2026.05 | 39.1 | — | 75.5 | — | 69.1 | 53.2 | 19.6 | — | |
| CA2C-SEMPublication=Ours2025.08 | 36.03 | 65.22 | — | 63.84 | — | — | — | — | |
| CA2CPublication=ICCV 20252025.08 | 33.56 | 63.56 | — | 61.16 | — | — | — | — | |
| UNICONPublication=CVPR 20222025.08 | 31.49 | 55.1 | — | 49.9 | — | — | — | — | |
| Jo-SRCPublication=CVPR 20212025.08 | 23.8 | 58.15 | — | 38.52 | — | — | — | — | |
| JoCoRPublication=CVPR 20202025.08 | 15.49 | 53.01 | — | 32.7 | — | — | — | — | |
| Co-teachingPublication=NeurIPS 20182025.08 | 15.15 | 43.73 | — | 28.35 | — | — | — | — | |
| AVGBackbone=ResNet-18, Pre-training=Pretrained on ImageNet, Number of seeds=32025.09 | — | 52.73 | 59.89 | 64.02 | 66.79 | 69.07 | 69.43 | — | |
| sUWABackbone=ResNet-18, Pre-training=Pretrained on ImageNet, Number of seeds=32025.09 | — | 60.85 | 63.29 | 64.81 | 66.2 | 67.9 | 68.35 | — | |
| UWABackbone=ResNet-18, Pre-training=Pretrained on ImageNet, Number of seeds=32025.09 | — | 59.82 | 62.09 | 63.43 | 64.91 | 66.76 | 67.67 | 69.95 |