Image Classification on ImageNet (Top-1/Top-5 Accuracy)
69.6Top-1 AccuracyReal-valued
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Real-valuedNetwork=ResNet-18, W/A=32/32, OPs (×10^8)=18.19, Training Protocol=two-stage training2026.05 | 69.6 | 89.2 | |
| SURGENetwork=ResNet-18, W/A=1/1, OPs (×10^8)=1.63, Training Protocol=two-stage training2026.05 | 66.7 | 86.7 | |
| RBONN*Network=ResNet-18, W/A=1/1, OPs (×10^8)=1.63, Training Protocol=two-stage training2026.05 | 66.5 | 86.7 | |
| ReCUNetwork=ResNet-18, W/A=1/1, OPs (×10^8)=1.63, Training Protocol=two-stage training2026.05 | 66.4 | 86.5 | |
| ReActNetNetwork=ResNet-18, W/A=1/1, OPs (×10^8)=1.63, Training Protocol=two-stage training2026.05 | 65.9 | — |