Neuron (boone)

java.lang.Object
- boone.BrainPart
- - boone.Neuron

All Implemented Interfaces:

Storable, java.lang.Cloneable, java.lang.Comparable

Direct Known Subclasses:

MaxPoolNeuron, SpikingNeuron
```
public class Neuron
extends BrainPart
implements java.lang.Comparable
```
Class representing a Neuron or Unit in the Neural Network. It is designed to be base class and default implementation at the same time.
NOTE: - the actual worker methods of links (adding, removing) are found in the Link class.
- for persistence, the default methods in the Boone class work.
- A Neuron doesn't contain a reference to the NeuralNet it is contained by.

Version:

$Id$

Author:

August Mayer, Helmut A. Mayer

Field Summary

Fields
Modifier and Type	Field and Description
`protected double`	`errorSignal` The error signal used for training.
`protected double`	`externalInput` the external input signal, meant for the input of input neurons.
`protected double`	`input` Current input signal.
`protected boolean`	`inputNeuron` Indicates an input neuron.
`protected int`	`layer` The layer number.
`protected double`	`linkInput` The sum of the inputs from the links to this neuron.
`protected java.util.ArrayList<Link>`	`links` All the neuron's links.
`protected double`	`output` The current neuron output.
`protected boolean`	`outputNeuron` Indicates an output neuron.
`protected Position`	`position` The neuron position.
`protected boolean`	`usingBias` Indicates usage of bias value.

Fields inherited from class boone.BrainPart
activationFn, name, NO_ID, weight

Constructor Summary

Constructors
Constructor and Description
`Neuron()` Creates a default Neuron using bias and Sigmoid activation.
`Neuron(boolean in, boolean out, boolean bias)` Creates a default Neuron with the given parameters.
`Neuron(Function function)` Creates a default Neuron using bias and the given activation.

Method Summary

All Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`void`	`addLinkInput(Link link)` Adds a link signal to the neuron's input.
`void`	`addToBias(double deltaBias)` Adds the given value to the bias value.
`void`	`calculate()` Propagate the neuron input to the neuron output.
`Neuron`	`clone()` Returns a semi-deep clone of this neuron.
`int`	`compareTo(java.lang.Object o)` Compares the neurons with respect to the layer number.
`void`	`fromXML(org.jdom2.Element element)` Reads attributes from XML tree.
`double`	`getBias()` Returns the bias value.
`double`	`getErrorSignal()` Returns the current error signal.
`double`	`getExternalInput()`
`double`	`getInput()` Return the neuron input, which is the sum of `externalInput`, `linkInput`, and `BrainPart.weight` (if `usingBias` is enabled).
`Link`	`getInputLink(int index)` Returns the input link of given index.
`int`	`getInputLinkCount()` Returns the number of input links.
`int`	`getLayer()` Returns the layer number of the neuron.
`Link`	`getLink(int index)` get a link
`int`	`getLinkCount()` Returns the number of links of this neuron.
`double`	`getLinkInput()` Return the value of the `linkInput`; please see above.
`java.util.ArrayList<Link>`	`getLinks()` Get all the input and output links.
`Link`	`getLinkTo(Neuron other)` Returns the link connecting this neuron to the given other neuron, which may also be the same neuron.
`double`	`getOutput()` Returns the current output value of this neuron.
`Link`	`getOutputLink(int index)` Returns the output link of given index.
`int`	`getOutputLinkCount()` Returns the number of output links.
`Position`	`getPosition()` Returns the position of the neuron.
`boolean`	`isHiddenNeuron()` Return true iff the neuron is neither input nor output neuron.
`boolean`	`isInputNeuron()` Indicates the input neuron state.
`boolean`	`isOutputNeuron()`
`boolean`	`isUsingBias()`
`void`	`propagateOutput()` Sends the neuron's activation onto the link.
`void`	`reset()` Resets internal neuron attributes.
`void`	`setBias(double bias)` Sets the bias value.
`void`	`setErrorSignal(double errorSignal)` Sets the error signal.
`void`	`setExternalInput(double externalInput)` Set the neuron's external input, e.g.
`void`	`setInput(double input)` Sets the current input value of the neuron.
`void`	`setInputNeuron(boolean inputNeuron)` define whether this neuron is an input neuron.
`void`	`setLayer(int layer)` Sets the layer number.
`void`	`setLinkInput(double linkInput)` Sets the neuron's link input.
`void`	`setOutput(double output)` Set the current output value of this neuron.
`void`	`setOutputNeuron(boolean outputNeuron)` define whether this neuron is an output neuron.
`void`	`setPosition(Position position)` Sets the neuron position.
`void`	`setUsingBias(boolean usingBias)`
`java.lang.String`	`toString()` Returns a string representation of the neuron.
`org.jdom2.Element`	`toXML(org.jdom2.Element element)` Writes attributes to XML tree.

Methods inherited from class boone.BrainPart
addGradient, addToWeight, getAccu, getActivationFn, getGradient, getID, getLastGradient, getName, getUniqueID, getWeight, isTrainable, randomize, randomize, setAccu, setActivationFn, setGradient, setID, setLastGradient, setName, setTrainable, setWeight

Methods inherited from class java.lang.Object
equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

- Field Detail
  - position
```
protected Position position
```
    The neuron position.
  - inputNeuron
```
protected boolean inputNeuron
```
    Indicates an input neuron.
  - outputNeuron
```
protected boolean outputNeuron
```
    Indicates an output neuron.
  - layer
```
protected int layer
```
    The layer number.
  - usingBias
```
protected boolean usingBias
```
    Indicates usage of bias value.
  - errorSignal
```
protected double errorSignal
```
    The error signal used for training.
  - externalInput
```
protected double externalInput
```
    the external input signal, meant for the input of input neurons.
  - linkInput
```
protected double linkInput
```
    The sum of the inputs from the links to this neuron. This is calculated step by step when the input link objects call the addLinkInput(boone.Link) method.
  - input
```
protected double input
```
    Current input signal. Calculated in the calculate() method by adding the linkInput and the externalInput values.
  - output
```
protected double output
```
    The current neuron output.
  - links
```
protected java.util.ArrayList<Link> links
```
    All the neuron's links.
- Constructor Detail
  - Neuron
```
public Neuron()
```
    Creates a default Neuron using bias and Sigmoid activation.
  - Neuron
```
public Neuron(Function function)
```
    Creates a default Neuron using bias and the given activation.
    
    Parameters:
    
    function - the activation function
  - Neuron
```
public Neuron(boolean in,
              boolean out,
              boolean bias)
```
    Creates a default Neuron with the given parameters. For input neurons the activation function to set to Identity.
    
    Parameters:
    
    in - an input neuron
    
    out - an output neuron
    
    bias - true, if using bias
- Method Detail
  - clone
```
public Neuron clone()
```
    Returns a semi-deep clone of this neuron. The clone's link list is empty and its ID is set to NO_ID. Note that the name is a reference to the cloned neuron.
    
    Overrides:
    
    clone in class BrainPart
    
    Returns:
    
    a semi-deep clone
  - compareTo
```
public int compareTo(java.lang.Object o)
```
    Compares the neurons with respect to the layer number.
    
    Specified by:
    
    compareTo in interface java.lang.Comparable
    
    Parameters:
    
    o - a neuron
    
    Returns:
    
    if o is larger -1, smaller 1, equal 0
  - reset
```
public void reset()
```
    Resets internal neuron attributes.
    
    Overrides:
    
    reset in class BrainPart
  - calculate
```
public void calculate()
```
    Propagate the neuron input to the neuron output.
    The neuron input (externalInput and linkInput) are already filled with the input values at this time.
    Doesn't calculate the output links; see propagateOutput() instead.
    
    Calculates the following values for this neuron: input, and output. Resets the linkInput to zero.
    
    Overrides:
    
    calculate in class BrainPart
  - propagateOutput
```
public void propagateOutput()
```
    Sends the neuron's activation onto the link.
    Calls Link.propagate() for each output link of this neuron.
  - isInputNeuron
```
public boolean isInputNeuron()
```
    Indicates the input neuron state. Note that a neuron may be both, input and output neuron.
    
    Returns:
    
    true, if input neuron
  - setInputNeuron
```
public void setInputNeuron(boolean inputNeuron)
```
    define whether this neuron is an input neuron.
  - isOutputNeuron
```
public boolean isOutputNeuron()
```
    Returns:
    
    true if we are an output neuron.
  - setOutputNeuron
```
public void setOutputNeuron(boolean outputNeuron)
```
    define whether this neuron is an output neuron.
  - isHiddenNeuron
```
public boolean isHiddenNeuron()
```
    Return true iff the neuron is neither input nor output neuron.
  - getPosition
```
public Position getPosition()
```
    Returns the position of the neuron.
    
    Returns:
    
    the position
  - setPosition
```
public void setPosition(Position position)
```
    Sets the neuron position.
    
    Parameters:
    
    position - the position
  - getLayer
```
public int getLayer()
```
    Returns the layer number of the neuron.
    
    Returns:
    
    the number, -1, if it is not decided
  - setLayer
```
public void setLayer(int layer)
```
    Sets the layer number. Only, use this method, if you are very sure what you are doing.
    
    Parameters:
    
    layer - the number
  - getExternalInput
```
public double getExternalInput()
```
    Returns:
    
    Value of property externalInput.
  - setExternalInput
```
public void setExternalInput(double externalInput)
```
    Set the neuron's external input, e.g. the environment signal for input neurons. This is the correct place to put the input pattern (but it is probably more convenient to use NeuralNet.setInput(double[]).
    
    Please also note that this attribute may be reset to zero on network evaluation, if the NeuralNet.setZeroInputAfterFirstCycle(boolean) flag is set to true (e.g. for Hopfield networks).
    
    Parameters:
    
    externalInput - the external input value
  - getInput
```
public double getInput()
```
    Return the neuron input, which is the sum of externalInput, linkInput, and BrainPart.weight (if usingBias is enabled). Calculated in the calculate() method.
    
    Returns:
    
    input value of this neuron
  - setInput
```
public void setInput(double input)
```
    Sets the current input value of the neuron. Please note that on calculate(), this value is calculated as the sum of externalInput, linkInput, and BrainPart.weight (the latter only if usingBias is enabled), and thus overwritten on network calculation.
    
    Parameters:
    
    input - New value of property input.
  - getOutput
```
public double getOutput()
```
    Returns the current output value of this neuron.
    
    Returns:
    
    the output value
  - setOutput
```
public void setOutput(double output)
```
    Set the current output value of this neuron. Only for special cases, or internal use.
    
    Parameters:
    
    output - New value of property output.
  - getBias
```
public double getBias()
```
    Returns the bias value.
    
    Returns:
    
    the bias
  - setBias
```
public void setBias(double bias)
```
    Sets the bias value.
    
    Parameters:
    
    bias - the bias
  - addToBias
```
public void addToBias(double deltaBias)
```
    Adds the given value to the bias value.
    
    Parameters:
    
    deltaBias - the bias change value
  - getErrorSignal
```
public double getErrorSignal()
```
    Returns the current error signal.
    
    Returns:
    
    an error signal
  - setErrorSignal
```
public void setErrorSignal(double errorSignal)
```
    Sets the error signal.
    
    Parameters:
    
    errorSignal - a value used for training
  - setLinkInput
```
public void setLinkInput(double linkInput)
```
    Sets the neuron's link input.
    
    Parameters:
    
    linkInput - the value
  - addLinkInput
```
public void addLinkInput(Link link)
```
    Adds a link signal to the neuron's input.
    
    Parameters:
    
    link - the link on which the signal is sent
  - getLinkInput
```
public double getLinkInput()
```
    Return the value of the linkInput; please see above.
  - isUsingBias
```
public boolean isUsingBias()
```
    Returns:
    
    true if this neuron should use the bias value.
  - setUsingBias
```
public void setUsingBias(boolean usingBias)
```
    Parameters:
    
    usingBias - whether the neuron should use the bias value.
  - getLinkCount
```
public int getLinkCount()
```
    Returns the number of links of this neuron.
    
    Returns:
    
    the total number of links
  - getLink
```
public Link getLink(int index)
```
    get a link
  - getLinks
```
public java.util.ArrayList<Link> getLinks()
```
    Get all the input and output links.
    
    Returns:
    
    a list of links
  - getLinkTo
```
public Link getLinkTo(Neuron other)
```
    Returns the link connecting this neuron to the given other neuron, which may also be the same neuron. This method also considers the direction of the connection. If this neuron has a directed connection to the other neuron, the link is returned. If it has the opposite direction, null is returned. Of course, with an directed connection, the link between A and B (if there) is always returned.
    
    Parameters:
    
    other - the other neuron possibly linked to this neuron
    
    Returns:
    
    the link from this to other (null, if no connection)
  - getInputLinkCount
```
public int getInputLinkCount()
```
    Returns the number of input links.
    
    Returns:
    
    the number of links which receive signals
  - getInputLink
```
public Link getInputLink(int index)
```
    Returns the input link of given index.
    
    Returns:
    
    the input link, null if not there
  - getOutputLinkCount
```
public int getOutputLinkCount()
```
    Returns the number of output links.
    
    Returns:
    
    the number of links which transmit the neuron's output
  - getOutputLink
```
public Link getOutputLink(int index)
```
    Returns the output link of given index.
    
    Returns:
    
    the output link, null if not there
  - fromXML
```
public void fromXML(org.jdom2.Element element)
```
    Reads attributes from XML tree.
    
    Specified by:
    
    fromXML in interface Storable
    
    Overrides:
    
    fromXML in class BrainPart
    
    Parameters:
    
    element - an XML element
  - toXML
```
public org.jdom2.Element toXML(org.jdom2.Element element)
```
    Writes attributes to XML tree.
    
    Specified by:
    
    toXML in interface Storable
    
    Overrides:
    
    toXML in class BrainPart
    
    Parameters:
    
    element - the parent element
    
    Returns:
    
    the neuron element
  - toString
```
public java.lang.String toString()
```
    Returns a string representation of the neuron.
    
    Overrides:
    
    toString in class BrainPart
    
    Returns:
    
    a string describing the neuron

Class Neuron

Field Summary

Fields inherited from class boone.BrainPart

Constructor Summary

Method Summary

Methods inherited from class boone.BrainPart

Methods inherited from class java.lang.Object

Field Detail

position

inputNeuron

outputNeuron

layer

usingBias

errorSignal

externalInput

linkInput

input

output

links

Constructor Detail

Neuron

Neuron

Neuron

Method Detail

clone

compareTo

reset

calculate

propagateOutput

isInputNeuron

setInputNeuron

isOutputNeuron

setOutputNeuron

isHiddenNeuron

getPosition

setPosition

getLayer

setLayer

getExternalInput

setExternalInput

getInput

setInput

getOutput

setOutput

getBias

setBias

addToBias

getErrorSignal

setErrorSignal

setLinkInput

addLinkInput

getLinkInput

isUsingBias

setUsingBias

getLinkCount

getLink

getLinks

getLinkTo

getInputLinkCount

getInputLink

getOutputLinkCount

getOutputLink

fromXML

toXML

toString