block_multivalued: Multivalued Block

In neuroimaginador/dl4ni: Deep Learning Flows for Neuroimaging

Description

This function is used to append a multivalued block to a previously existing object.

Usage

block_multivalued(object, hidden_layers = list(), num_values = 2,
  units = 1)

Arguments

object	(keras object) The object where to append the block.
hidden_layers	(list) List of layers, with types and corresponding parameters.
num_values	(integer) The number of different possible values in the output of the block.
units	(integer) Number of output units of the block.
hidden_activation	(character) Activation function to use in the hidden layers of the block.
hidden_dropout	(numeric in [0, 1]) The dropout to use in the hidden layers.
params	(list) List of parameters to apply, if not listed in the previous ones.

Details

A multivalued block consists on units different paths, with the common input object, each path is composed by a set of hidden layers (all with the same architecture defined in the arguments of this function), and ended by a dense layer with num_values units and "sigmoid" activation followed by another dense layer with one unit and "linear" activation.

The final output is the concatenation of these units independent paths.

Value

The object composed with the multivalued block.

neuroimaginador/dl4ni documentation built on May 3, 2019, 5:47 p.m.