Novel regularization by shuffling channels of feature maps.
Python
Shuffle Channel Layer
Novel regularization by shuffling channels of feature maps.
The order of channels of input tensor for convolution layer is fixed.
This might make training more diffcult and enable model to learn more useful information.