org.joone.util
Class LearningSwitch

java.lang.Object
  org.joone.engine.Synapse
      org.joone.io.StreamInputSynapse
          org.joone.io.InputSwitchSynapse
              org.joone.util.LearningSwitch

All Implemented Interfaces:

java.io.Serializable, InputPatternListener, Learnable, LearnableSynapse, NeuralElement, OutputPatternListener, Inspectable, InputSynapse, PlugInListener

public class LearningSwitch
extends InputSwitchSynapse

This class is useful to switch the input data set of a neural network from a training set to a validation set depending on the 'validation' parameter contained in the Monitor object. Very useful during a training phase to test the generalization capacity of the neural network with a validation data set never seen before.

Author:

pmarrone

See Also:

Serialized Form

Field Summary

Fields inherited from class org.joone.io.InputSwitchSynapse

inputs

Fields inherited from class org.joone.io.StreamInputSynapse

cols, currentRow, EOF, InputVector, plugInListeners, tokens

Fields inherited from class org.joone.engine.Synapse

array, b_pattern, bitems, bouts, count, enabled, fwdLock, inps, items, learnable, m_batch, m_pattern, myLearner, outs, revLock

Constructor Summary

LearningSwitch()
Creates a new instance of LearningSwitch

Method Summary

boolean	addTrainingSet(StreamInputSynapse tSet)
boolean	addValidationSet(StreamInputSynapse vSet)
java.util.TreeSet	check() Check that parameters are set correctly.
Pattern	fwdGet() Connects the right input synapse depending on the Monitor's 'validation' parameter
Pattern	fwdGet(InputConnector conn) Connects the right input synapse depending on the Monitor's 'validation' parameter.
StreamInputSynapse	getTrainingSet() Getter for property trainingSet.
StreamInputSynapse	getValidationSet() Getter for property validationSet.
void	removeTrainingSet()
void	removeValidationSet()
void	setTrainingSet(StreamInputSynapse trainingSet) Setter for property trainingSet.
void	setValidationSet(StreamInputSynapse validationSet) Setter for property validationSet.

Methods inherited from class org.joone.io.InputSwitchSynapse

addInputSynapse, backward, dataChanged, forward, getActiveInput, getActiveSynapse, getAdvancedColumnSelector, getAllInputs, getDefaultInput, getDefaultSynapse, getFirstRow, getInputSynapse, getInspections, getLastRow, getMonitor, getName, getOutputDimension, getPlugIn, gotoFirstLine, gotoLine, init, initInputStream, isBuffered, isEOF, isStepCounter, numColumns, readAll, removeAllInputs, removeInputSynapse, reset, resetInput, resetSwitch, revPut, setActiveInput, setActiveSynapse, setAllInputs, setBuffered, setDecimalPoint, setDefaultInput, setDefaultSynapse, setMonitor, setName, setOutputDimension, setStepCounter

Methods inherited from class org.joone.io.StreamInputSynapse

addPlugIn, addPlugInListener, fireDataChanged, getAllPlugInListeners, getCurrentRow, getDecimalPoint, getFirstCol, getInputPatterns, getInputVector, getLastCol, getMaxBufSize, getStream, getTokens, handleExceptionWithError, handleExceptionWithWarn, InspectableTitle, Inspections, readObjectBase, removeAllPlugIns, removePlugInListener, restart, setAdvancedColumnSelector, setArrays, setColList, setDimensions, setEOF, setFirstCol, setFirstRow, setInputPatterns, setLastCol, setLastRow, setMaxBufSize, setPlugin, setPlugIn, setTokens, writeObjectBase, zeroPattern

Methods inherited from class org.joone.engine.Synapse

addNoise, canCountSteps, fwdPattern_consume, fwdPattern_produce, fwdPut, getCount, getFwdLock, getIgnoreBefore, getInputDimension, getLearner, getLearningRate, getMomentum, getRevLock, getSynapseNameSafely, getWeights, initLearner, isEnabled, isInputFull, isLoopBack, isOutputFull, randomize, readResolve, revGet, revPattern_consume, revPattern_produce, setEnabled, setIgnoreBefore, setInputDimension, setInputFull, setLearningRate, setLoopBack, setMomentum, setOutputFull, setWeights, warnLogger, warnLogger

Methods inherited from class java.lang.Object

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

Constructor Detail

LearningSwitch

public LearningSwitch()

Creates a new instance of LearningSwitch

Method Detail

addTrainingSet

public boolean addTrainingSet(StreamInputSynapse tSet)

addValidationSet

public boolean addValidationSet(StreamInputSynapse vSet)

removeTrainingSet

public void removeTrainingSet()

removeValidationSet

public void removeValidationSet()

fwdGet

public Pattern fwdGet()

Connects the right input synapse depending on the Monitor's 'validation' parameter

Specified by:

fwdGet in interface InputPatternListener

Overrides:

fwdGet in class InputSwitchSynapse

Returns:

neural.engine.Pattern

fwdGet

public Pattern fwdGet(InputConnector conn)

Connects the right input synapse depending on the Monitor's 'validation' parameter. Added to be compatible with the InputConnector class.

Overrides:

fwdGet in class InputSwitchSynapse

Returns:

neural.engine.Pattern

check

public java.util.TreeSet check()

Description copied from class: InputSwitchSynapse

Check that parameters are set correctly.

Specified by:

check in interface NeuralElement

Overrides:

check in class InputSwitchSynapse

Returns:

validation errors.

See Also:

Synapse

getTrainingSet

public StreamInputSynapse getTrainingSet()

Getter for property trainingSet. Added for XML serialization

Returns:

Value of property trainingSet.

setTrainingSet

public void setTrainingSet(StreamInputSynapse trainingSet)

Setter for property trainingSet. Added for XML serialization

Parameters:

trainingSet - New value of property trainingSet.

getValidationSet

public StreamInputSynapse getValidationSet()

Getter for property validationSet. Added for XML serialization

Returns:

Value of property validationSet.

setValidationSet

public void setValidationSet(StreamInputSynapse validationSet)

Setter for property validationSet. Added for XML serialization

Parameters:

validationSet - New value of property validationSet.