org.simnet.neurons
Class LMSNeuron

java.lang.Object
  org.simnet.interfaces.Neuron
      org.simnet.neurons.LMSNeuron

All Implemented Interfaces:

GaugeSource

public class LMSNeuron
extends Neuron

LMS Neuron. A way of implementing the delta rule (AKA Widrow Hoff) rule in a single neuron. It is assumed that one of the synapses attaching to this neuron is a signal synapse, which carries the target value used in learning.

Field Summary

Fields inherited from class org.simnet.interfaces.Neuron

activation, fanIn, fanOut, lowerBound, upperBound

Constructor Summary

LMSNeuron()
Default constructor needed for external calls which create neurons then set their parameters.

LMSNeuron(Neuron n)
This constructor is used when creating a neuron of one type from another neuron of another type Only values common to different types of neuron are copied.

Method Summary

Neuron	duplicate() Returns a duplicate ClampedNeuron (used, e.g., in copy/paste).
double	getLearningRate()
static String	getName()
int	getTimeType()
double	getWeightedInputs() Overrides superclass implementation by ignoring inputs from targetValue synapses.
void	setLearningRate(double learningRate) Sets the learning rate.
void	update() Update neuron.

Methods inherited from class org.simnet.interfaces.Neuron

addActivation, addSource, addTarget, checkBounds, clear, clip, connectedToWeight, containsString, debug, decrementActivation, deleteConnectedSynapses, deleteFanIn, deleteFanOut, duplicate, getActivation, getAverageInput, getBuffer, getFanIn, getFanOut, getGaugeValue, getHelp, getId, getIncrement, getInputValue, getLowerBound, getMotorCoupling, getNeuronTypeIndex, getNumberOfActiveInputs, getParentNetwork, getSensoryCoupling, getSummedIncomingWeights, getToolTipText, getTotalInput, getType, getTypeList, getUpperBound, getX, getY, incrementActivation, initCastor, isInput, isOutput, randomize, randomizeBuffer, round, setActivation, setBuffer, setFanIn, setFanOut, setId, setIncrement, setInputValue, setLowerBound, setMotorCoupling, setParentNetwork, setSensoryCoupling, setTypeList, setUpperBound, setX, setY, toString, updateConnectedOutward

Methods inherited from class java.lang.Object

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

Constructor Detail

LMSNeuron

public LMSNeuron()

Default constructor needed for external calls which create neurons then set their parameters.

LMSNeuron

public LMSNeuron(Neuron n)

This constructor is used when creating a neuron of one type from another neuron of another type Only values common to different types of neuron are copied.

Parameters:

n - Neuron to make the type

Method Detail

getTimeType

public int getTimeType()

Specified by:

getTimeType in class Neuron

Returns:

time type.

duplicate

public Neuron duplicate()

Returns a duplicate ClampedNeuron (used, e.g., in copy/paste).

Specified by:

duplicate in class Neuron

Returns:

Duplicated neuron

update

public void update()

Update neuron.

Specified by:

update in class Neuron

getWeightedInputs

public double getWeightedInputs()

Overrides superclass implementation by ignoring inputs from targetValue synapses.

Overrides:

getWeightedInputs in class Neuron

Returns:

weighted input to this node

getName

public static String getName()

Returns:

Name of neuron type.

getLearningRate

public double getLearningRate()

Returns:

Learning rate.

setLearningRate

public void setLearningRate(double learningRate)

Sets the learning rate.

Parameters:

learningRate - Learning rate to set