Image Classification on CIFAR-10 (ACCm, p-value)
91.33Mean AccuracySimCLR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SimCLRPre-training Task=ImageNet, Model Condition=Clean2026.04 | 91.33 | 1 | |
| ArmSSL (SimCLR)Pre-training Task=ImageNet, Model Condition=Watermarked2026.04 | 89.83 | 0 | |
| MOCO v2Pre-training Task=ImageNet, Model Condition=Clean2026.04 | 88.13 | 1 | |
| ArmSSL (MOCO v2)Pre-training Task=ImageNet, Model Condition=Watermarked2026.04 | 87.97 | 0 | |
| SimCLRPre-training Task=CIFAR-10, Model Condition=Clean2026.04 | 86.9 | 1 | |
| ArmSSL (SimCLR)Pre-training Task=CIFAR-10, Model Condition=Watermarked2026.04 | 86.6 | 0 | |
| BYOLPre-training Task=CIFAR-10, Model Condition=Clean2026.04 | 86.52 | 1 | |
| SimSiamPre-training Task=ImageNet, Model Condition=Clean2026.04 | 86.42 | 1 | |
| ArmSSL (SimSiam)Pre-training Task=ImageNet, Model Condition=Watermarked2026.04 | 86.31 | 0 | |
| ArmSSL (BYOL)Pre-training Task=CIFAR-10, Model Condition=Watermarked2026.04 | 86.11 | 0 | |
| MOCO v2Pre-training Task=CIFAR-10, Model Condition=Clean2026.04 | 86.1 | 1 | |
| SimSiamPre-training Task=CIFAR-10, Model Condition=Clean2026.04 | 86.1 | 1 | |
| ArmSSL (MOCO v2)Pre-training Task=CIFAR-10, Model Condition=Watermarked2026.04 | 85.84 | 0 | |
| ArmSSL (SimSiam)Pre-training Task=CIFAR-10, Model Condition=Watermarked2026.04 | 85.83 | 0 | |
| BYOLPre-training Task=Imagenette, Model Condition=Clean2026.04 | 70.6 | 0.98 | |
| SimCLRPre-training Task=Imagenette, Model Condition=Clean2026.04 | 69.9 | 0.89 | |
| ArmSSL (SimCLR)Pre-training Task=Imagenette, Model Condition=Watermarked2026.04 | 69.8 | 0 | |
| ArmSSL (BYOL)Pre-training Task=Imagenette, Model Condition=Watermarked2026.04 | 67.4 | 0 |