Normalization
Switchable Normalization
Introduced by Luo et al. in Differentiable Learning-to-Normalize via Switchable NormalizationSwitchable Normalization combines three types of statistics estimated channel-wise, layer-wise, and minibatch-wise by using instance normalization, layer normalization, and batch normalization respectively. Switchable Normalization switches among them by learning their importance weights.
Source: Differentiable Learning-to-Normalize via Switchable Normalization
Papers
| Paper | Code | Results | Date | Stars |
|---|
Tasks
| Task | Papers | Share |
|---|---|---|
| Pedestrian Detection | 2 | 18.18% |
| Retinal Vessel Segmentation | 1 | 9.09% |
| Multi-Task Learning | 1 | 9.09% |
| Person Re-Identification | 1 | 9.09% |
| Person Search | 1 | 9.09% |
| Image Generation | 1 | 9.09% |
| Ischemic Stroke Lesion Segmentation | 1 | 9.09% |
| Lesion Segmentation | 1 | 9.09% |
| Semantic Segmentation | 1 | 9.09% |
Usage Over Time
Components
| Component | Type |
|
|---|---|---|
Batch Normalization
|
Normalization | |
Instance Normalization
|
Normalization | |
Layer Normalization
|
Normalization | |
Softmax
|
Output Functions |
