Image Classification on CIFAR-10+ (test)
3.44Test Error Rate (%)EAS (DenseNet on C10+)
Evaluation Results
| Method | Links | |
|---|---|---|
| EAS (DenseNet on C10+)Depth=76, Params=10.7M2017.07 | 3.44 | |
| DenseNet-BC (L = 190, k = 40)Depth=190, Params=25.6M2017.07 | 3.46 | |
| DenseNet-BCGrowth Rate (k)=40, Depth=190, Params=25.6M2016.08 | 3.46 | |
| DenseNet-BC (L = 250, k = 24)Depth=250, Params=15.3M2017.07 | 3.62 | |
| DenseNet-BCGrowth Rate (k)=24, Depth=250, Params=15.3M2016.08 | 3.62 | |
| NAS (post-processing)Depth=39, Params=37.4M2017.07 | 3.65 | |
| DenseNet (L = 100, k = 24)Depth=100, Params=27.2M2017.07 | 3.74 | |
| DenseNetGrowth Rate (k)=24, Depth=100, Params=27.2M2016.08 | 3.74 | |
| DenseNetGrowth Rate (k)=12, Depth=100, Params=7.0M2016.08 | 4.1 | |
| Wide ResNetDepth=28, Params=36.5M2016.08 | 4.17 | |
| DenseNet-BCGrowth Rate (k)=12, Depth=100, Params=0.8M2016.08 | 4.51 | |
| FractalNet with Dropout/Drop-pathDepth=21, Params=38.6M2016.08 | 4.6 | |
| ResNet (pre-activation)Depth=1001, Params=10.2M2016.08 | 4.62 | |
| Wide ResNetDepth=16, Params=11.0M2016.08 | 4.81 | |
| ResNet with Stochastic DepthDepth=1202, Params=10.2M2016.08 | 4.91 | |
| FractalNetDepth=21, Params=38.6M2016.08 | 5.22 | |
| ResNet with Stochastic DepthDepth=110, Params=1.7M2016.08 | 5.23 | |
| DenseNetGrowth Rate (k)=12, Depth=40, Params=1.0M2016.08 | 5.24 | |
| ResNet (pre-activation)Depth=164, Params=1.7M2016.08 | 5.46 | |
| ResNet (reported by [13])Depth=110, Params=1.7M2016.08 | 6.41 | |
| ResNetDepth=110, Params=1.7M2016.08 | 6.61 | |
| All-CNNDepth=-, Params=-2016.08 | 7.25 | |
| Highway NetworkDepth=-, Params=-2016.08 | 7.72 | |
| Deeply Supervised NetDepth=-, Params=-2016.08 | 7.97 | |
| Network in NetworkDepth=-, Params=-2016.08 | 8.81 |