Package libai.nn.supervised.backpropagation

Class StandardBackpropagation

java.lang.Object

libai.nn.supervised.backpropagation.StandardBackpropagation

All Implemented Interfaces:

Backpropagation

Direct Known Subclasses:

MomentumBackpropagation, ResilientBackpropagation

public class StandardBackpropagation
extends java.lang.Object
implements Backpropagation

Created by kronenthaler on 08/01/2017.

Field Summary

Fields

Modifier and Type	Field	Description
protected Column[]	b
protected Column[]	d
protected Function[]	func
protected int	layers
protected Matrix[]	M
protected NeuralNetwork	nn
protected int[]	nperlayer
protected Column[]	u
protected Matrix[]	W
protected Column[]	Y
protected Row[]	Yt

Constructor Summary

Constructors

Constructor	Description
StandardBackpropagation()

Method Summary

Modifier and Type	Method	Description
void	initialize(NeuralNetwork nn, int[] nperlayer, Function[] functions, Matrix[] W, Column[] Y, Column[] b, Column[] u)	Initializes the Backpropagation algorithm with the basic structures needed.
void	train(Column[] patterns, Column[] answers, double alpha, int epochs, int offset, int length, double minerror)	Trains a neural network using the backpropagation implementaion.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

nn

protected NeuralNetwork nn

nperlayer

protected int[] nperlayer

layers

protected int layers

func

protected Function[] func

W

protected Matrix[] W

Y

protected Column[] Y

b

protected Column[] b

u

protected Column[] u

d

protected Column[] d

Yt

protected Row[] Yt

M

protected Matrix[] M

Constructor Detail

StandardBackpropagation

public StandardBackpropagation()

Method Detail

initialize

public void initialize(NeuralNetwork nn,
                       int[] nperlayer,
                       Function[] functions,
                       Matrix[] W,
                       Column[] Y,
                       Column[] b,
                       Column[] u)

Description copied from interface: Backpropagation

Initializes the Backpropagation algorithm with the basic structures needed. This method usually should be called in the NeuralNetwork's train method right before calling the Backpropagation's train method.

Specified by:

initialize in interface Backpropagation

train

public void train(Column[] patterns,
                  Column[] answers,
                  double alpha,
                  int epochs,
                  int offset,
                  int length,
                  double minerror)

Description copied from interface: Backpropagation

Trains a neural network using the backpropagation implementaion.

Specified by:

train in interface Backpropagation