Classification on PneumoniaMNIST
98.28AccuracyDUCS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DUCSBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 98.28 | — | |
| DUCSBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 97.71 | — | |
| EVABackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 97.66 | — | |
| EVABackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 97.46 | — | |
| DUCSBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 97.33 | — | |
| Full datasetBackbone=ResNet-182026.03 | 97.27 | — | |
| EntropyBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 97.07 | — | |
| EL2NBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 97.07 | — | |
| EVABackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 97.07 | — | |
| AUMBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 96.68 | — | |
| RandomBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 96.48 | — | |
| ForgettingBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 96.48 | — | |
| AUMBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 96.29 | — | |
| AUMBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 96.29 | — | |
| MedViT-SResolution=2242025.07 | 96.1 | 99.5 | |
| EntropyBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 96.09 | — | |
| RandomBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 95.9 | — | |
| EL2NBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 95.9 | — | |
| RandomBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 95.7 | — | |
| ForgettingBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 95.7 | — | |
| EntropyBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 95.7 | — | |
| ForgettingBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 95.52 | — | |
| ModerateBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 95.33 | — | |
| Google AutoML Vision2026.02 | 94.6 | 99.1 | |
| EL2NBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 94.34 | — | |
| EVABackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 94.34 | — | |
| ModerateBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 94.23 | — | |
| TDDSBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 94.18 | — | |
| DUCSBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 93.32 | — | |
| CCSBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 92.97 | — | |
| RandomBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 92.77 | — | |
| MAIL-RN18Base network=ResNet182026.02 | 92.49 | 98.83 | |
| MAIL-M2Base network=MobileNetV22026.02 | 92.05 | 98.6 | |
| DRIFA-Net2026.02 | 91.83 | 98.24 | |
| TDDSBackbone=ResNet-18, Selection Ratio (beta)=50%2026.03 | 91.67 | — | |
| MFMSA2026.02 | 91.49 | 98.15 | |
| Gloria2026.02 | 91.25 | 97.88 | |
| M3Att2026.02 | 91.25 | 97.82 | |
| CCSBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 91.21 | — | |
| VICRegBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 91.03 | — | |
| CCSBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 91.02 | — | |
| CCSBackbone=ResNet-18, Selection Ratio (beta)=70%2026.03 | 91.02 | — | |
| BYOLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 90.99 | — | |
| ModerateBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 90.82 | — | |
| MTTU-Net2026.02 | 90.47 | 97.79 | |
| DDA-Net2026.02 | 90.25 | 97.25 | |
| TDDSBackbone=ResNet-18, Selection Ratio (beta)=30%2026.03 | 90.02 | — | |
| NAT2026.02 | 89.95 | 97.12 | |
| AMIMV-SSLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 89.9 | — | |
| POTTER2026.02 | 89.89 | 97.74 | |
| MuMu2026.02 | 89.84 | 97.75 | |
| HAMLET2026.02 | 89.42 | 97.65 | |
| SymmFlowResolution=282025.07 | 89.4 | 91.6 | |
| LatMSFResolution=2242025.07 | 89.4 | 94.4 | |
| SimCLRBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 88.81 | — | |
| ResNet-50Resolution=2242026.02 | 88.4 | 96.2 | |
| ResNet-50Resolution=2242025.07 | 88.4 | 96.2 | |
| NNCLRBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 88.24 | — | |
| ModerateBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 88.14 | — | |
| Barlow TwinsBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 88.08 | — | |
| MSFResolution=282025.07 | 88 | 95.2 | |
| AutoKeras2026.02 | 87.8 | 94.7 | |
| AutoKerasResolution=282025.07 | 87.8 | 94.7 | |
| MoCoV3Backbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 87.34 | — | |
| DINOBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 87.24 | — | |
| ReSSLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 87.21 | — | |
| ResNet-18Resolution=2242026.02 | 86.4 | 95.6 | |
| ResNet-18Resolution=2242025.07 | 86.4 | 95.6 | |
| auto-sklearn2026.02 | 85.5 | 94.2 | |
| auto-sklearnResolution=282025.07 | 85.5 | 94.2 | |
| ResNet-18Resolution=282026.02 | 85.4 | 94.4 | |
| ResNet-50Resolution=282026.02 | 85.4 | 94.8 | |
| ResNet-18Resolution=282025.07 | 85.4 | 94.4 | |
| ResNet-50Resolution=282025.07 | 85.4 | 94.8 | |
| TDDSBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 82.85 | — | |
| G-LoG bi-filtration (MLP)Sigma (σ)=0.52026.02 | 82.5 | 91 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=12026.02 | 82.5 | 90.6 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=02026.02 | 81.7 | 90.7 | |
| BioVLMTrain Params=30K2026.04 | 79.65 | — | |
| G-LoG bi-filtration (MLP)Sigma (σ)=1.52026.02 | 79.5 | 89.1 | |
| Topo-Med (MLP)2026.02 | 76.2 | 84.5 | |
| CoCoOpTrain Params=44K2026.04 | 76.01 | — | |
| PromptSRCTrain Params=69K2026.04 | 75.06 | — | |
| MaPLeTrain Params=5.3M2026.04 | 74.81 | — | |
| KgCoOpTrain Params=3K2026.04 | 73.24 | — | |
| CoOpTrain Params=3K2026.04 | 72.7 | — | |
| TCPTrain Params=0.5M2026.04 | 72.48 | — | |
| EntropyBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 71.48 | — | |
| BiomedCoOpTrain Params=3K2026.04 | 70.51 | — | |
| BioMedCLIPTrain Params=02026.04 | 59.46 | — | |
| ProGradTrain Params=3K2026.04 | 57.64 | — | |
| ForgettingBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 55.86 | — | |
| AUMBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 38.87 | — | |
| EL2NBackbone=ResNet-18, Selection Ratio (beta)=10%2026.03 | 26.17 | — | |
| AMIMV-SSLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 98.19 | |
| Barlow TwinsBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 92.38 | |
| BYOLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 97.45 | |
| DINOBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 90.29 | |
| MoCoV3Backbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 98.39 | |
| NNCLRBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 94.27 |