Classification on RetinaMNIST
70.69ACCMAIL-RN18
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MAIL-RN18Base network=ResNet182026.02 | 70.69 | 81.32 | |
| DINOv2-LVD142MBackbone=ViT-L/14, Training Scale=large scale datasets, Evaluation protocol=Linear probe2026.06 | 69.5 | — | |
| MAIL-M2Base network=MobileNetV22026.02 | 69.45 | 80.54 | |
| DRIFA-Net2026.02 | 69.38 | 79.81 | |
| MFMSA2026.02 | 69.1 | 78.84 | |
| DDA-Net2026.02 | 67.62 | 78.46 | |
| Gloria2026.02 | 66.89 | 77.55 | |
| NAT2026.02 | 66.58 | 77.6 | |
| VISReg-Inet22KBackbone=ViT-L/14, Training Scale=large scale datasets, Evaluation protocol=Linear probe2026.06 | 66.5 | — | |
| I-JEPABackbone=ViT-H/14, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 65.75 | — | |
| MoCoV3Backbone=ViT-B/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 64.5 | — | |
| MuMu2026.02 | 64.28 | 77.41 | |
| VISRegBackbone=ViT-L/14, Heuristics=Excluded, Evaluation protocol=Linear probe2026.06 | 64.25 | — | |
| data2vecBackbone=ViT-L/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 63.75 | — | |
| iBOTBackbone=ViT-B/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 63.75 | — | |
| POTTER2026.02 | 63.54 | 74.26 | |
| HAMLET2026.02 | 63.38 | 75.81 | |
| M3Att2026.02 | 63.15 | 76.9 | |
| iBOTBackbone=ViT-L/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 63 | — | |
| MAEBackbone=ViT-L/16, Heuristics=Excluded, Evaluation protocol=Linear probe2026.06 | 63 | — | |
| DINOBackbone=ViT-B/16, Heuristics=Included, Evaluation protocol=Linear probe2026.06 | 62.5 | — | |
| VISRegBackbone=ViT-B/16, Heuristics=Excluded, Evaluation protocol=Linear probe2026.06 | 62.25 | — | |
| MTTU-Net2026.02 | 61.75 | 75.59 | |
| MedMamba-XResolution=224x2242026.03 | 57 | 71.9 | |
| HamClsResolution=224x2242026.03 | 56.75 | 76.24 | |
| AMIMV-SSLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 56.75 | — | |
| MedViT-SResolution=2242025.07 | 56.1 | 77.3 | |
| LatMSFResolution=2242025.07 | 54 | 78.8 | |
| MedViT-SResolution=224x2242026.03 | 53.4 | 75.2 | |
| Google AutoML Vision2026.02 | 53.1 | 75 | |
| ResNet-50Resolution=282026.02 | 52.8 | 72.6 | |
| ResNet-50Resolution=282025.07 | 52.8 | 72.6 | |
| Barlow TwinsBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 52.5 | — | |
| ResNet-18Resolution=282026.02 | 52.4 | 71.7 | |
| ResNet-18Resolution=282025.07 | 52.4 | 71.7 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=02026.02 | 51.8 | 61.1 | |
| auto-sklearn2026.02 | 51.5 | 69 | |
| auto-sklearnResolution=282025.07 | 51.5 | 69 | |
| MSFResolution=282025.07 | 51.4 | 73.1 | |
| ResNet-50Resolution=2242026.02 | 51.1 | 71.6 | |
| ResNet-50Resolution=2242025.07 | 51.1 | 71.6 | |
| SymmFlowResolution=282025.07 | 50.7 | 70.2 | |
| AutoKeras2026.02 | 50.3 | 71.9 | |
| AutoKerasResolution=282025.07 | 50.3 | 71.9 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=0.52026.02 | 50 | 62.4 | |
| G-LoG bi-filtration (MLP)Sigma (σ)=12026.02 | 49.8 | 64.3 | |
| ResNet-18Resolution=2242026.02 | 49.3 | 71 | |
| ResNet-18Resolution=2242025.07 | 49.3 | 71 | |
| VICRegBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 49.15 | — | |
| G-LoG bi-filtration (MLP)Sigma (σ)=1.52026.02 | 48.3 | 63.8 | |
| MoCoV3Backbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 47.75 | — | |
| NNCLRBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 47.4 | — | |
| SimCLRBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 46.45 | — | |
| DINOBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 46.2 | — | |
| Topo-Med (MLP)2026.02 | 45.8 | 72.8 | |
| ReSSLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 43.15 | — | |
| BYOLBackbone=ResNet-50, Initialization=Random, Training Strategy=Self-supervised2026.04 | 41.2 | — | |
| AMIMV-SSLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 76.02 | |
| Barlow TwinsBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 67.19 | |
| BYOLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 51.64 | |
| DINOBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 58.72 | |
| MoCoV3Backbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 65.37 | |
| NNCLRBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 62.72 | |
| ReSSLBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 60.35 | |
| SimCLRBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 63.38 | |
| VICRegBackbone=ResNet-50, Initialization=random, Training Protocol=self-supervised2026.04 | — | 63.49 |