Image Classification on ImageNet (Top-1 Error, Top-5 Error)
21.36Top-1 ErrorIAM-D
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| IAM-DBackbone=ResNet-50, Epochs=200, Batch Size=1024, Initial Learning Rate=0.2, ρ=0.12026.05 | 21.36 | 5.7 | |
| IAM-SBackbone=ResNet-50, Epochs=200, Batch Size=1024, Initial Learning Rate=0.2, ρ=0.22026.05 | 21.72 | 5.9 | |
| SAMBackbone=ResNet-50, Epochs=200, Batch Size=1024, Initial Learning Rate=0.2, ρ=0.052026.05 | 21.8 | 5.99 | |
| SGDBackbone=ResNet-50, Epochs=200, Batch Size=1024, Initial Learning Rate=0.22026.05 | 22.66 | 6.51 |