Classification on STL10 (test)
90.1Robust Accuracy (RA)Net2Vec
Evaluation results
| Method | Links | |
|---|---|---|
| Net2VecBackbone=ResNet502024.11 | 90.1 | |
| NAVESetting=N34, Backbone=ResNet502024.11 | 89.5 | |
| NAVESetting=N4, Backbone=ResNet502024.11 | 88.7 | |
| CRAFTBackbone=ResNet502024.11 | 88.6 | |
| ResNet50Backbone=ResNet502024.11 | 87.6 | |
| NAVESetting=N234, Backbone=ResNet502024.11 | 86.2 | |
| ZePAD (W-MSE)Pre-training Dataset (Dp)=CIFAR10, SSL Encoder=W-MSE, Backbone=ResNet-182026.02 | 70.28 | |
| ZePAD (BYOL)Pre-training Dataset (Dp)=ImageNet, SSL Encoder=BYOL, Backbone=ResNet-182026.02 | 67.8 | |
| W-MSE (Baseline)Pre-training Dataset (Dp)=CIFAR10, SSL Method=W-MSE, Backbone=ResNet-182026.02 | 41.16 | |
| BYOL (Baseline)Pre-training Dataset (Dp)=ImageNet, SSL Method=BYOL, Backbone=ResNet-182026.02 | 33.02 | |
| CONE-SHAPBackbone=ResNet502024.11 | 10.4 |