Activation Normalization is a type of normalization used for flow-based generative models; specifically it was introduced in the GLOW architecture. An ActNorm layer performs an affine transformation of the activations using a scale and bias parameter per channel, similar to batch normalization. These parameters are initialized such that the post-actnorm activations per-channel have zero mean and unit variance given an initial minibatch of data. This is a form of data dependent initilization. After initialization, the scale and bias are treated as regular trainable parameters that are independent of the data.
Source: Glow: Generative Flow with Invertible 1x1 ConvolutionsPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Image Dehazing | 2 | 11.11% |
Image Enhancement | 2 | 11.11% |
Pseudo Label | 1 | 5.56% |
Offline RL | 1 | 5.56% |
Benchmarking | 1 | 5.56% |
Molecular Docking | 1 | 5.56% |
Pose Prediction | 1 | 5.56% |
Low-Light Image Enhancement | 1 | 5.56% |
Zero-Shot Learning | 1 | 5.56% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |