weka.classifiers.functions.neural

Class NeuralConnection

java.lang.Object

weka.classifiers.functions.neural.NeuralConnection

All Implemented Interfaces:

java.io.Serializable, RevisionHandler

Direct Known Subclasses:

NeuralNode

public abstract class NeuralConnection
extends java.lang.Object
implements java.io.Serializable, RevisionHandler

Abstract unit in a NeuralNetwork.

Version:

$Revision: 8034 $

Author:

Malcolm Ware (mfw4@cs.waikato.ac.nz)

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static int	CONNECTED This flag is set once the unit has a connection.
static int	INPUT This unit is an input unit.
static int	OUTPUT This unit is an output unit.
static int	PURE_INPUT This unit is a pure input unit.
static int	PURE_OUTPUT This unit is a pure output unit.
static int	UNCONNECTED This unit is not connected to any others.

Constructor Summary

Constructors

Constructor and Description
NeuralConnection(java.lang.String id) Constructs The unit with the basic connection information prepared for use.

Method Summary

Modifier and Type	Method and Description
static boolean	connect(NeuralConnection s, NeuralConnection t) Connects two units together.
static boolean	disconnect(NeuralConnection s, NeuralConnection t) Disconnects two units.
void	drawHighlight(java.awt.Graphics g, int w, int h) Call this function to draw the node highlighted.
void	drawInputLines(java.awt.Graphics g, int w, int h) Call this function to draw the nodes input connections.
void	drawNode(java.awt.Graphics g, int w, int h) Call this function to draw the node.
void	drawOutputLines(java.awt.Graphics g, int w, int h) Call this function to draw the nodes output connections.
abstract double	errorValue(boolean calculate) Call this to get the error value of this unit.
java.lang.String	getId()
int[]	getInputNums() Use this to get easy access to the input numbers.
NeuralConnection[]	getInputs() Use this to get easy access to the inputs.
int	getNumInputs()
int	getNumOutputs()
int[]	getOutputNums() Use this to get easy access to the output numbers.
NeuralConnection[]	getOutputs() Use this to get easy access to the outputs.
int	getType()
double	getX()
double	getY()
boolean	onUnit(java.awt.Graphics g, int x, int y, int w, int h) Call this function to determine if the point at x,y is on the unit.
abstract double	outputValue(boolean calculate) Call this to get the output value of this unit.
void	removeAllInputs() This function will remove all the inputs to this unit.
void	removeAllOutputs() This function will remove all outputs to this unit.
abstract void	reset() Call this to reset the unit for another run.
abstract void	restoreWeights() Call this to have the connection restore from the saved weights.
abstract void	saveWeights() Call this to have the connection save the current weights.
void	setType(int t)
void	setX(double x)
void	setY(double y)
void	updateWeights(double l, double m) Call this function to update the weight values at this unit.
double	weightValue(int n) Call this to get the weight value on a particular connection.

Methods inherited from class java.lang.Object

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

Methods inherited from interface weka.core.RevisionHandler

getRevision

Field Detail

UNCONNECTED

public static final int UNCONNECTED

This unit is not connected to any others.

See Also:

Constant Field Values

PURE_INPUT

public static final int PURE_INPUT

This unit is a pure input unit.

See Also:

Constant Field Values

PURE_OUTPUT

public static final int PURE_OUTPUT

This unit is a pure output unit.

See Also:

Constant Field Values

INPUT

public static final int INPUT

This unit is an input unit.

See Also:

Constant Field Values

OUTPUT

public static final int OUTPUT

This unit is an output unit.

See Also:

Constant Field Values

CONNECTED

public static final int CONNECTED

This flag is set once the unit has a connection.

See Also:

Constant Field Values

Constructor Detail

NeuralConnection

public NeuralConnection(java.lang.String id)

Constructs The unit with the basic connection information prepared for use.

Parameters:

id - the unique id of the unit

Method Detail

getId

public java.lang.String getId()

Returns:

The identity string of this unit.

getType

public int getType()

Returns:

The type of this unit.

setType

public void setType(int t)

Parameters:

t - The new type of this unit.

reset

public abstract void reset()

Call this to reset the unit for another run. It is expected by that this unit will call the reset functions of all input units to it. It is also expected that this will not be done if the unit has already been reset (or atleast appears to be).

outputValue

public abstract double outputValue(boolean calculate)

Call this to get the output value of this unit.

Parameters:

calculate - True if the value should be calculated if it hasn't been already.

Returns:

The output value, or NaN, if the value has not been calculated.

errorValue

public abstract double errorValue(boolean calculate)

Call this to get the error value of this unit.

Parameters:

calculate - True if the value should be calculated if it hasn't been already.

Returns:

The error value, or NaN, if the value has not been calculated.

saveWeights

public abstract void saveWeights()

Call this to have the connection save the current weights.

restoreWeights

public abstract void restoreWeights()

Call this to have the connection restore from the saved weights.

weightValue

public double weightValue(int n)

Call this to get the weight value on a particular connection.

Parameters:

n - The connection number to get the weight for, -1 if The threshold weight should be returned.

Returns:

This function will default to return 1. If overridden, it should return the value for the specified connection or if -1 then it should return the threshold value. If no value exists for the specified connection, NaN will be returned.

updateWeights

public void updateWeights(double l,
                          double m)

Call this function to update the weight values at this unit. After the weights have been updated at this unit, All the input connections will then be called from this to have their weights updated.

Parameters:

l - The learning Rate to use.

m - The momentum to use.

getInputs

public NeuralConnection[] getInputs()

Use this to get easy access to the inputs. It is not advised to change the entries in this list (use the connecting and disconnecting functions to do that)

Returns:

The inputs list.

getOutputs

public NeuralConnection[] getOutputs()

Use this to get easy access to the outputs. It is not advised to change the entries in this list (use the connecting and disconnecting functions to do that)

Returns:

The outputs list.

getInputNums

public int[] getInputNums()

Use this to get easy access to the input numbers. It is not advised to change the entries in this list (use the connecting and disconnecting functions to do that)

Returns:

The input nums list.

getOutputNums

public int[] getOutputNums()

Use this to get easy access to the output numbers. It is not advised to change the entries in this list (use the connecting and disconnecting functions to do that)

Returns:

The outputs list.

getX

public double getX()

Returns:

the x coord.

getY

public double getY()

Returns:

the y coord.

setX

public void setX(double x)

Parameters:

x - The new value for it's x pos.

setY

public void setY(double y)

Parameters:

y - The new value for it's y pos.

onUnit

public boolean onUnit(java.awt.Graphics g,
                      int x,
                      int y,
                      int w,
                      int h)

Call this function to determine if the point at x,y is on the unit.

Parameters:

g - The graphics context for font size info.

x - The x coord.

y - The y coord.

w - The width of the display.

h - The height of the display.

Returns:

True if the point is on the unit, false otherwise.

drawNode

public void drawNode(java.awt.Graphics g,
                     int w,
                     int h)

Call this function to draw the node.

Parameters:

g - The graphics context.

w - The width of the drawing area.

h - The height of the drawing area.

drawHighlight

public void drawHighlight(java.awt.Graphics g,
                          int w,
                          int h)

Call this function to draw the node highlighted.

Parameters:

g - The graphics context.

w - The width of the drawing area.

h - The height of the drawing area.

drawInputLines

public void drawInputLines(java.awt.Graphics g,
                           int w,
                           int h)

Call this function to draw the nodes input connections.

Parameters:

g - The graphics context.

w - The width of the drawing area.

h - The height of the drawing area.

drawOutputLines

public void drawOutputLines(java.awt.Graphics g,
                            int w,
                            int h)

Call this function to draw the nodes output connections.

Parameters:

g - The graphics context.

w - The width of the drawing area.

h - The height of the drawing area.

removeAllInputs

public void removeAllInputs()

This function will remove all the inputs to this unit. In doing so it will also terminate the connections at the other end.

removeAllOutputs

public void removeAllOutputs()

This function will remove all outputs to this unit. In doing so it will also terminate the connections at the other end.

getNumInputs

public int getNumInputs()

Returns:

The number of input connections.

getNumOutputs

public int getNumOutputs()

Returns:

The number of output connections.

connect

public static boolean connect(NeuralConnection s,
                              NeuralConnection t)

Connects two units together.

Parameters:

s - The source unit.

t - The target unit.

Returns:

True if the units were connected, false otherwise.

disconnect

public static boolean disconnect(NeuralConnection s,
                                 NeuralConnection t)

Disconnects two units.

Parameters:

s - The source unit.

t - The target unit.

Returns:

True if the units were disconnected, false if they weren't (probably due to there being no connection).