Image Classification on SVHN (test) (Error Rate, FLOPs, and Parameter Count)
1.71Error RateFSNP
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FSNPCNN=ResNet-1642025.02 | 1.71 | 59.1 | 37.4 | |
| STNPFCNN=ResNet-1642025.02 | 1.74 | 63 | — | |
| FSNPCNN=DenseNet-402025.02 | 1.78 | 50.2 | 56.8 | |
| BaselineCNN=ResNet-1642025.02 | 1.78 | — | — | |
| hbfp8_16Backbone=DN-40, Mantissa bits=8, Weight storage bits=162018.04 | 1.79 | — | — | |
| DSCCNN=DenseNet-402025.02 | 1.8 | 45.1 | 54.9 | |
| fp32Backbone=DN-402018.04 | 1.8 | — | — | |
| SLIMCNN=DenseNet-402025.02 | 1.81 | 49.8 | 56.6 | |
| SLIMCNN=ResNet-1642025.02 | 1.81 | 54.9 | 34.3 | |
| hbfp12_16Backbone=DN-40, Mantissa bits=12, Weight storage bits=162018.04 | 1.85 | — | — | |
| BaselineCNN=DenseNet-402025.02 | 1.89 | — | — | |
| fp32Backbone=RN-502018.04 | 1.89 | — | — | |
| hbfp12_16Backbone=WRN-16-8, Mantissa bits=12, Weight storage bits=162018.04 | 1.94 | — | — | |
| hbfp12_16Backbone=RN-50, Mantissa bits=12, Weight storage bits=162018.04 | 1.96 | — | — | |
| hbfp8_16Backbone=RN-50, Mantissa bits=8, Weight storage bits=162018.04 | 1.98 | — | — | |
| hbfp8_16Backbone=WRN-16-8, Mantissa bits=8, Weight storage bits=162018.04 | 1.98 | — | — | |
| fp32Backbone=WRN-16-82018.04 | 2 | — | — | |
| IAM-D2026.05 | 3.13 | — | — | |
| IAM-S2026.05 | 3.13 | — | — | |
| ASAM2026.05 | 3.24 | — | — | |
| NPRWCNN=ResNet-1642025.02 | 3.37 | 58.9 | 37.1 | |
| SAM2026.05 | 3.47 | — | — | |
| SGD2026.05 | 3.82 | — | — |