class BatchNorm : Batch normalization layer. [Ioffe & Szegedy, 2015].

class Concat : Layer to concat tensors according to the given axis.

class Conv2d : 2d convolution layer.

class Dense : Fully connected layer.

class Elementwise : Layer to combine inputs by applying element-wise operation.

class Flatten : Layer to reshape input into a matrix.

class GlobalMaxPool2d : 2d global max pooling layer.

class GlobalMeanPool2d : 2d global mean pooling layer.

class MaxPool2d : 2d max pooling layer.

class MeanPool2d : 2d mean pooling layer.

class Layer : The base layer class.

class LayerList : Layer to stack a group of layers.

class Relu : Layer to apply the rectified linear unit. [Nair & Hinton, 2010].

class Reshape : Layer to change the dimensions of input.

class Transpose : Layer to permute the dimensions of input.


Input(…) : Create a placeholder as input.