org.simnet.networks
Class Backprop

java.lang.Object
  org.simnet.interfaces.Network
      org.simnet.networks.Backprop

All Implemented Interfaces:

EventListener, WorldListener

Direct Known Subclasses:

Elman

public class Backprop
extends Network

Backprop implements a standard three layer backpropagation network.

Field Summary

Fields inherited from class org.simnet.interfaces.Network

CONTINUOUS, DISCRETE

Constructor Summary

Backprop()
Default constructor.

Backprop(int inputs, int hidden, int outputs, Layout layout)
Construct a backprop network with a specified number of input, hidden, and output layers.

Method Summary

void	attachInputsAndOutputs() Attach training files to SNARLI network.
void	batchIterate() Batch train for one iteration.
void	batchTrain() Forwards to Snarli batchTrain() method.
protected void	buildInitialNetwork() Build the default network.
void	buildSnarliNetwork() Create the Snarli network.
void	defaultInit() Build network and initialize nodes and weights to appropriate values.
double[]	getBiases(StandardNetwork net) Returns bias values.
protected SigmoidalNeuron	getDefaultNeuron() Return the default neuron, with settings, for backprop nets.
int	getEpochs()
double	getError()
int	getErrorInterval()
double	getEta()
edu.wlu.cs.levy.SNARLI.BPLayer	getHid()
edu.wlu.cs.levy.SNARLI.BPLayer	getInp()
double	getMu()
int	getNHidden()
int	getNInputs()
int	getNOutputs()
edu.wlu.cs.levy.SNARLI.BPLayer	getOut()
File	getTrainingINFile()
double[][]	getTrainingInputs()
File	getTrainingOUTFile()
double[][]	getTrainingOutputs()
void	iterate() Iterate network training.
void	randomize() Randomize the network.
void	setBiases(StandardNetwork net, double[] biases) Set bias values for all neurons in this network.
void	setEpochs(int epochs)
void	setError(double error)
void	setErrorInterval(int errorInterval)
void	setEta(double eta)
void	setHid(edu.wlu.cs.levy.SNARLI.BPLayer hid)
void	setInp(edu.wlu.cs.levy.SNARLI.BPLayer inp)
void	setMu(double mu)
void	setNHidden(int nHidden)
void	setNInputs(int nInputs)
void	setNOutputs(int nOutputs)
void	setOut(edu.wlu.cs.levy.SNARLI.BPLayer out)
void	setTrainingINFile(File trainingINFile) Sets the input training file.
void	setTrainingInputs(double[][] trainingInputs)
void	setTrainingOUTFile(File trainingOUTFile) Sets the output training file.
void	setTrainingOutputs(double[][] trainingOutputs)
void	train() Train the network.
void	update() The core update function of the neural network.
void	updateSimbrainNetwork() Update connections and biases of simbrain network.

Methods inherited from class org.simnet.interfaces.Network

addNetwork, addNetworkList, addNetworkListener, addNeuron, addNeuron, addNeuronList, addWeight, addWeight, changeNeuron, changeSynapse, checkAllBounds, clearActivations, clearInputs, close, deleteNetwork, deleteNeuron, deleteNeuron, deleteWeight, deleteWeight, fireClampChanged, fireCouplingChanged, fireNetworkChanged, fireNeuronAdded, fireNeuronChanged, fireNeuronDeleted, fireSubnetAdded, fireSubnetDeleted, fireSynapseAdded, fireSynapseChanged, fireSynapseDeleted, getClampNeurons, getClampWeights, getCouplingList, getDepth, getFlatNetworkList, getFlatNeuronList, getFlatSynapseList, getId, getIndents, getInputNeurons, getInteractionMode, getListenerList, getNetwork, getNetworkList, getNetworkParent, getNetworkThread, getNeuron, getNeuron, getNeuronCount, getNeuronList, getOutputNeurons, getPrecision, getRoot, getRoundingOff, getState, getSynapse, getTime, getTimeLabel, getTimeStep, getTimeType, getType, getUnits, getWeight, getWeight, getWeight, getWeightCount, getWeightList, getWorkspace, init, initNeurons, initParents, initWeights, isRoot, isRoundOffActivationValues, isUpdateCompleted, randomizeNeurons, randomizeWeights, removeNetworkListener, round, roundAll, setClampNeurons, setClampWeights, setId, setInteractionMode, setLowerBounds, setNetworkList, setNetworkParent, setNetworkThread, setNeuronList, setPrecision, setRoundingOff, setRoundOffActivationValues, setTime, setTimeStep, setUpdateCompleted, setUpperBounds, setWeightList, setWeightsToZero, setWorkspace, toString, updateAllNetworks, updateAllNeurons, updateAllWeights, updateIds, updateInputs, updateTime, updateTimeType, updateTopLevel, updateWorldListeners, updateWorlds, worldChanged

Methods inherited from class java.lang.Object

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

Constructor Detail

Backprop

public Backprop()

Default constructor.

Backprop

public Backprop(int inputs,
                int hidden,
                int outputs,
                Layout layout)

Construct a backprop network with a specified number of input, hidden, and output layers.

Parameters:

inputs - Number of neurons to be inputs

hidden - Number of neurons to be hidden

outputs - Number of neurons to be outputs

layout - Defines how the neurons are layed out

Method Detail

defaultInit

public void defaultInit()

Build network and initialize nodes and weights to appropriate values.

buildInitialNetwork

protected void buildInitialNetwork()

Build the default network.

buildSnarliNetwork

public void buildSnarliNetwork()

Create the Snarli network.

getDefaultNeuron

protected SigmoidalNeuron getDefaultNeuron()

Return the default neuron, with settings, for backprop nets.

Returns:

the neuron, with appropriate settings, that should be used in building a backprop net.

update

public void update()

The core update function of the neural network. Calls the current update function on each neuron, decays all the neurons, and checks their bounds.

Specified by:

update in class Network

train

public void train()

Train the network.

iterate

public void iterate()

Iterate network training.

attachInputsAndOutputs

public void attachInputsAndOutputs()

Attach training files to SNARLI network.

updateSimbrainNetwork

public void updateSimbrainNetwork()

Update connections and biases of simbrain network.

batchTrain

public void batchTrain()

Forwards to Snarli batchTrain() method.

batchIterate

public void batchIterate()

Batch train for one iteration.

randomize

public void randomize()

Randomize the network.

getEpochs

public int getEpochs()

Returns:

Returns the epochs.

setEpochs

public void setEpochs(int epochs)

Parameters:

epochs - The epochs to set.

getError

public double getError()

Returns:

Returns the error.

setError

public void setError(double error)

Parameters:

error - The error to set.

getErrorInterval

public int getErrorInterval()

Returns:

Returns the error_interval.

setErrorInterval

public void setErrorInterval(int errorInterval)

Parameters:

errorInterval - The errorInterval to set.

getEta

public double getEta()

Returns:

Returns the eta.

setEta

public void setEta(double eta)

Parameters:

eta - The eta to set.

getHid

public edu.wlu.cs.levy.SNARLI.BPLayer getHid()

Returns:

Returns the hid.

setHid

public void setHid(edu.wlu.cs.levy.SNARLI.BPLayer hid)

Parameters:

hid - The hid to set.

getInp

public edu.wlu.cs.levy.SNARLI.BPLayer getInp()

Returns:

Returns the inp.

setInp

public void setInp(edu.wlu.cs.levy.SNARLI.BPLayer inp)

Parameters:

inp - The inp to set.

getMu

public double getMu()

Returns:

Returns the mu.

setMu

public void setMu(double mu)

Parameters:

mu - The mu to set.

getNHidden

public int getNHidden()

Returns:

Returns the n_hidden.

setNHidden

public void setNHidden(int nHidden)

Parameters:

nHidden - The nHidden to set.

getNInputs

public int getNInputs()

Returns:

Returns the n_inputs.

setNInputs

public void setNInputs(int nInputs)

Parameters:

nInputs - The nInputs to set.

getNOutputs

public int getNOutputs()

Returns:

Returns the n_outputs.

setNOutputs

public void setNOutputs(int nOutputs)

Parameters:

nOutputs - The nOutputs to set.

getOut

public edu.wlu.cs.levy.SNARLI.BPLayer getOut()

Returns:

Returns the out.

setOut

public void setOut(edu.wlu.cs.levy.SNARLI.BPLayer out)

Parameters:

out - The out to set.

getTrainingInputs

public double[][] getTrainingInputs()

Returns:

Returns the training_inputs.

setTrainingInputs

public void setTrainingInputs(double[][] trainingInputs)

Parameters:

trainingInputs - The trainingInputs to set.

getTrainingOutputs

public double[][] getTrainingOutputs()

Returns:

Returns the trainingOutputs.

setTrainingOutputs

public void setTrainingOutputs(double[][] trainingOutputs)

Parameters:

trainingOutputs - The trainingOutputs to set.

getBiases

public double[] getBiases(StandardNetwork net)

Returns bias values.

Parameters:

net - Network

Returns:

Neuron biases

setBiases

public void setBiases(StandardNetwork net,
                      double[] biases)

Set bias values for all neurons in this network.

Parameters:

biases - Array of new bias values

net - Network to get biases

getTrainingINFile

public File getTrainingINFile()

Returns:

Returns the input training file.

setTrainingINFile

public void setTrainingINFile(File trainingINFile)

Sets the input training file.

Parameters:

trainingINFile - File to set input training

getTrainingOUTFile

public File getTrainingOUTFile()

Returns:

Returns the output training file.

setTrainingOUTFile

public void setTrainingOUTFile(File trainingOUTFile)

Sets the output training file.

Parameters:

trainingOUTFile - File to set output training