jas.ai.ann
Class InputNodeLayer

java.lang.Object
  jas.ai.ann.NodeLayer
      jas.ai.ann.InputNodeLayer

public class InputNodeLayer
extends NodeLayer

The input node layer is used to store values of the input layer of the neural network. It is able to manage the scaling of the values, according to the vectors of minimun and maximum values. The layer is equipped with a monitor, which checks that stored values are within the allowed range. When a particular node goes out of bounds the method isOutOfScale() returns true.

Title: JAS - Neural network package

Description: Java Agent-based Simulation library

Copyright (C) 2002-3 Michele Sonnessa

This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2.1 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307, USA.

Author:

Michele Sonnessa

Constructor Summary

InputNodeLayer(int nodes, boolean withBias)
Build an input node layer with a given number of nodes.

InputNodeLayer(int nodes, boolean withBias, double[] minValues, double[] maxValues)
Build an input node layer with a given number of nodes and given ranges for each node values.

Method Summary

double	getMaxValue(int nodeIndex) Return the upper bound for the given node index.
double[]	getMaxVector() Return a vector of maximum value for the input nodes.
Jama.Matrix	getMinMaxMatrix() Return a Jama.Matrix of minimum and maximum values for the input nodes.
double	getMinValue(int nodeIndex) Return the lower bound for the given node index.
double[]	getMinVector() Return a vector of minimum value for the input nodes.
double	getUnscaledNodeValue(int nodeIndex) Return the current value for the given node index, using the orginal scale.
double[]	getUnscaledVector() Return the current value for each node of the layer, using the orginal scale.
boolean	isOutOfScale() Report if setUnscaledNodeValue method recevied values out of bounds.
void	resetOutOfScale() Reset the out of scale monitor.
void	setMaxValue(int nodeIndex, double value) Set the new upper bound for the given node index.
void	setMinMaxValue(int nodeIndex, double minValue, double maxValue) Set the new lower and upper bounds for the given node index.
void	setMinValue(int nodeIndex, double value) Set the new lower bound for the given node index.
void	setUnscaledNodeValue(int nodeIndex, double value) Set the new value for the given node index.
void	setUnscaledVector(double[] values) Set node values in the original scale for each cell of the vector.

Methods inherited from class jas.ai.ann.NodeLayer

getArray, getArrayCopy, getBias, getNodesNumber, getNodeValue, getRowVector, getTrasposedVector, getValue, getVectorLength, hasBias, setBias, setNodeValue, setValue, setVector, toString

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

InputNodeLayer

public InputNodeLayer(int nodes,
                      boolean withBias)

Build an input node layer with a given number of nodes. WARNING MinMax vector is initialized with 0 in min and 1 in max

Parameters:

nodes - The number of nodes in the layers.

withBias - If true the position 0 of the vector will contain the bias and the total number of cells in the vector will be [nodes + 1].

InputNodeLayer

public InputNodeLayer(int nodes,
                      boolean withBias,
                      double[] minValues,
                      double[] maxValues)

Build an input node layer with a given number of nodes and given ranges for each node values.

Parameters:

nodes - The number of nodes in the layers.

withBias - If true the position 0 of the vector will contain the bias and the total number of cells in the vector will be [nodes + 1].

minValues - A vector of [nodes] values containing the lower bound for each node.

maxValues - A vector of [nodes] values containing the upper bound for each node.

Method Detail

getMaxValue

public double getMaxValue(int nodeIndex)

Return the upper bound for the given node index.

Parameters:

nodeIndex - The index of the node. WARNING: The index bounds are not checked!

Returns:

The maximum value for given index node.

getMaxVector

public double[] getMaxVector()

Return a vector of maximum value for the input nodes.

Returns:

The max vector.

getMinMaxMatrix

public Jama.Matrix getMinMaxMatrix()

Return a Jama.Matrix of minimum and maximum values for the input nodes. The row 0 contains min values and the row 1 the max ones.

Returns:

The min-max matrix.

getMinValue

public double getMinValue(int nodeIndex)

Return the lower bound for the given node index.

Parameters:

nodeIndex - The index of the node. WARNING: The index bounds are not checked!

Returns:

The minimum value for given index node.

getMinVector

public double[] getMinVector()

Return a vector of minimum value for the input nodes.

Returns:

The min vector.

getUnscaledNodeValue

public double getUnscaledNodeValue(int nodeIndex)

Return the current value for the given node index, using the orginal scale. The scaling is within the range [0,1] according the current bound values. The scaling formula is scaledValue = currentValue * (max - min) + min.

Parameters:

The - index of the node. WARNING: The index bounds are not checked!

Returns:

The scaled value.

getUnscaledVector

public double[] getUnscaledVector()

Return the current value for each node of the layer, using the orginal scale. The scaling is within the range [0,1] according the current bound values. The scaling formula is scaledValue = currentValue * (max - min) + min.

Returns:

A double[] array with original scaled values. It is a copy of the internal vector: changes will not affect original values.

isOutOfScale

public boolean isOutOfScale()

Report if setUnscaledNodeValue method recevied values out of bounds.

resetOutOfScale

public void resetOutOfScale()

Reset the out of scale monitor.

setMaxValue

public void setMaxValue(int nodeIndex,
                        double value)

Set the new upper bound for the given node index.

Parameters:

nodeIndex - The index of the node. WARNING: The index bounds are not checked!

value - The new bound value.

setMinMaxValue

public void setMinMaxValue(int nodeIndex,
                           double minValue,
                           double maxValue)

Set the new lower and upper bounds for the given node index.

Parameters:

nodeIndex - The index of the node. WARNING: The index bounds are not checked!

minValue - The new lower bound value.

maxValue - The new upper bound value.

setMinValue

public void setMinValue(int nodeIndex,
                        double value)

Set the new lower bound for the given node index.

Parameters:

nodeIndex - The index of the node. WARNING: The index bounds are not checked!

value - The new bound value.

setUnscaledNodeValue

public void setUnscaledNodeValue(int nodeIndex,
                                 double value)

Set the new value for the given node index. The value stored in the vector is scaled using Min-Max range. The scaling is within the range [0,1] according the current bound values. The scaling formula is scaledValue = currentValue * (max - min) + min.

Parameters:

The - index of the node. WARNING: The index bounds are not checked!

setUnscaledVector

public void setUnscaledVector(double[] values)

Set node values in the original scale for each cell of the vector. If bias is enabled it is ignored.

Parameters:

values - A double[] array containig unscaled node values.