init.neuron {AMORE}R Documentation

Neuron constructor.

Description

Creates a neuron according to the structure estaablished by the AMORE package standard.

Usage

init.neuron(id,type,activation.function,output.links, output.aims, input.links, weights, bias, former.weight.change, former.bias.change, learning.rate, sum.delta.x, sum.delta.bias, momentum, delta)

Arguments

id Numerical index of the neuron (so as to be refered in a network operation).
type Either hidden or ouput, according to the layer the neuron belongs to.
activation.function The name of the characteristic function of the neuron. It can be "pureline", "tansig", "sigmoid" or even "custom" in case that the user wants to configure its own activation function accordingly defining f0 and f1.
output.links The id's of the neurons that accept the output value of this neuron as an input.
output.aims The location of the output of the neuron in the input set of the addressed neuron. Gives answer to: Is this output the first, the second, the third, ..., input at the addressed neuron?. Similarly for an output neuron: Is this output the first, the second, the third, ..., element of the output vector?
input.links The id's of the neurons whose outputs work as inputs for this neuron. Positive values represent that we take the outputs of other neurons as inputs. Negative values represent the coordinates of the input vector to be considered as inputs.
weights The multiplying factors of the input values.
former.weight.change
bias
former.bias.change Last increment in the bias parameter. Used by the momentum training technique.
learning.rate Learning rate parameter. Notice that we can use a different rate for each neuron.
sum.delta.x
sum.delta.bias Used as an acumulator of the changes to apply to the bias parameters in the batch training.
momentum Momentum constant used in the backpropagation with momentum learning criterium.
delta Used in the backpropagation method.

Value

init.neuron returns a single neuron. Mainly used to create a neural network object.

Author(s)

Manuel Castejón Limas. manuel.castejon@unileon.es
Joaquin Ordieres Meré. joaquin.ordieres@dim.unirioja.es
Ana González Marcos. ana.gonzalez@unileon.es
Alpha V. Pernía Espinoza. alpha.pernia@alum.unirioja.es
Eliseo P. Vergara González. eliseo.vergara@dim.unirioja.es
Francisco Javier Martinez de Pisón. francisco.martinez@dim.unirioja.es
Fernando Alba Elías. fernando.alba@unavarra.es

See Also

newff, random.init.NeuralNet, random.init.neuron, select.activation.function , init.neuron


[Package AMORE version 0.1.1 Index]