niftynet.layer.loss_autoencoder module

class niftynet.layer.loss_autoencoder.LossFunction(loss_type='VariationalLowerBound', loss_func_params={}, name='loss_function')

Bases: niftynet.layer.base_layer.Layer

layer_op(prediction)

make_callable_loss_func(type_str)

niftynet.layer.loss_autoencoder.variational_lower_bound(prediction)

This is the variational lower bound derived in Auto-Encoding Variational Bayes, Kingma & Welling, 2014 :param posterior_means: predicted means for the posterior :param posterior_logvar: predicted log variances for the posterior :param data_means: predicted mean parameter

for the voxels modelled as Gaussians

Parameters:

• data_logvar – predicted log variance parameter for the voxels modelled as Gaussians
• originals – the original inputs

Returns: