vae.autoencoder
===============

.. py:module:: vae.autoencoder


Classes
-------

.. autoapisummary::

   vae.autoencoder.KLDivergenceLayer
   vae.autoencoder.VAE


Functions
---------

.. autoapisummary::

   vae.autoencoder.sample_point_from_normal_distribution


Module Contents
---------------

.. py:function:: sample_point_from_normal_distribution(args)

   Sampling function using the reparameterization trick.


.. py:class:: KLDivergenceLayer

   Bases: :py:obj:`tensorflow.keras.layers.Layer`


   Custom layer to compute and add KL divergence to the model's loss. 


   .. py:method:: call(inputs)


.. py:class:: VAE(input_shape, conv_filters, conv_kernels, conv_strides, latent_space_dim)

   VAE represents a Deep Convolutional variational autoencoder architecture
   with mirrored encoder and decoder components.


   .. py:attribute:: input_shape


   .. py:attribute:: conv_filters


   .. py:attribute:: conv_kernels


   .. py:attribute:: conv_strides


   .. py:attribute:: latent_space_dim


   .. py:attribute:: reconstruction_loss_weight
      :value: 10000



   .. py:attribute:: encoder
      :value: None



   .. py:attribute:: decoder
      :value: None



   .. py:attribute:: model
      :value: None



   .. py:attribute:: _num_conv_layers


   .. py:attribute:: _shape_before_bottleneck
      :value: None



   .. py:attribute:: _model_input
      :value: None



   .. py:method:: summary()


   .. py:method:: compile(learning_rate=0.0001)


   .. py:method:: train(x_train, batch_size, num_epochs)


   .. py:method:: save(save_folder='.')


   .. py:method:: load_weights(weights_path)


   .. py:method:: reconstruct(images)


   .. py:method:: load(save_folder='.')
      :classmethod:



   .. py:method:: _calculate_combined_loss(y_target, y_predicted)


   .. py:method:: _calculate_reconstruction_loss(y_target, y_predicted)


   .. py:method:: _create_folder_if_it_doesnt_exist(folder)


   .. py:method:: _save_parameters(save_folder)


   .. py:method:: _save_weights(save_folder)


   .. py:method:: _build()


   .. py:method:: _build_autoencoder()


   .. py:method:: _build_decoder()


   .. py:method:: _add_decoder_input()


   .. py:method:: _add_dense_layer(decoder_input)


   .. py:method:: _add_reshape_layer(dense_layer)


   .. py:method:: _add_conv_transpose_layers(x)


   .. py:method:: _add_conv_transpose_layer(layer_index, x)


   .. py:method:: _add_decoder_output(x)


   .. py:method:: _build_encoder()


   .. py:method:: _add_encoder_input()


   .. py:method:: _add_conv_layers(encoder_input)


   .. py:method:: _add_conv_layer(layer_index, x)


   .. py:method:: _add_bottleneck(x)