input - input layer classes

The file yann.layers.input.py contains the definition for the input lyer modules.

class yann.layers.input.dropout_input_layer(mini_batch_size, id, x, dropout_rate=0.5, height=28, width=28, channels=1, mean_subtract=False, rng=None, verbose=2)[source]

Creates a new input_layer. The layer doesn’t do much except to take the networks’ x and reshapes it into images. This is needed because yann dataset module assumes data in vectorized image formats as used in mnist - theano tutorials.

This class also creates a branch between a mean subtracted and non-mean subtracted input. It always assumes as default to use the non-mean subtracted input but if mean_subtract flag is provided, it will use the other option.

Parameters:
  • x – theano.tensor variable with rows are vectorized images. if None, will create a new one.
  • mini_batch_size – Number of images in the data variable.
  • height – Height of each image.
  • width – Width of each image.
  • channels – Number of channels in each image.
  • mean_subtract – Defauly is False.
  • verbose – Similar to all of the toolbox.

Notes

Use input_layer.output to continue onwards with the network input_layer.output_shape will tell you the output size.

class yann.layers.input.dropout_tensor_layer(id, input, input_shape, rng=None, dropout_rate=0.5, verbose=2)[source]

This converts a theano tensor or a shared value into a layer. Simply the value becomes the layer’s outptus.

Parameters:
  • input – some tensor
  • input_shape – shape of the tensor
  • dropout_rate – default is 0.5, typically.
  • rng – Random number generator
  • verbose – Similar to all of the toolbox.

Notes

Use input_layer.output to continue onwards with the network input_layer.output_shape will tell you the output size.

class yann.layers.input.input_layer(mini_batch_size, x, id=-1, height=28, width=28, channels=1, mean_subtract=False, verbose=2)[source]

reshapes it into images. This is needed because yann dataset module assumes data in vectorized image formats as used in mnist - theano tutorials.

This class also creates a branch between a mean subtracted and non-mean subtracted input. It always assumes as default to use the non-mean subtracted input but if mean_subtract flag is provided, it will use the other option.

Parameters:
  • x – theano.tensor variable with rows are vectorized images.
  • y – theano.tensor variable
  • one_hot_y – theano.tensor variable
  • mini_batch_size – Number of images in the data variable.
  • height – Height of each image.
  • width – Width of each image.
  • id – Supply a layer id
  • channels – Number of channels in each image.
  • mean_subtract – Defauly is False.
  • verbose – Similar to all of the toolbox.

Notes

Use input_layer.output to continue onwards with the network input_layer.output_shape will tell you the output size. Use input_layer.x, input_layer.y and input_layer_one_hot_y tensors for connections.

class yann.layers.input.tensor_layer(id, input, input_shape, verbose=2)[source]

This converts a theano tensor or a shared value into a layer. Simply the value becomes the layer’s outptus.

Parameters:
  • input – some tensor
  • input_shape – shape of the tensor
  • verbose – Similar to all of the toolbox.

Notes

Use input_layer.output to continue onwards with the network input_layer.output_shape will tell you the output size.