wGmdh.jGmdh.oldskul
Class TwoInputModel

java.lang.Object
  wGmdh.jGmdh.oldskul.Node
      wGmdh.jGmdh.oldskul.Model
          wGmdh.jGmdh.oldskul.TwoInputModel

All Implemented Interfaces:

java.io.Serializable, java.lang.Comparable<Node>

Direct Known Subclasses:

ErrorPropagatingModel

public class TwoInputModel
extends Model

Polynomial GMDH node with two inputs, by L2 fitting

See Also:

Serialized Form

Nested Class Summary

class

TwoInputModel.VisitedHt Uses a Hashtable to store pairs of Nodes and inputs for which they have calculated the output, to avoid multiple calculations of the same thing.

Nested classes/interfaces inherited from class wGmdh.jGmdh.oldskul.Model

Model.Summand

Nested classes/interfaces inherited from class wGmdh.jGmdh.oldskul.Node

Node.Visited

Field Summary

TwoInputModel.VisitedHt

visitinfo

Fields inherited from class wGmdh.jGmdh.oldskul.Model

coeffs, errorMeasure, links, sortedLinkIds, structureSelectionCriterion

Fields inherited from class wGmdh.jGmdh.oldskul.Node

graphics, trainSetOutput, validationSetOutput, visited

Constructor Summary

TwoInputModel(java.util.ArrayList<double[]> regressionGoal, Performance selectionCriterion, Performance errorMeasure, Node inputL, Node inputR)

TwoInputModel(double[] regressionGoals, Performance selectionCriterion, Performance errorMeasure, Node... links)

TwoInputModel(Node... links)

TwoInputModel(Performance selectionCriterion, Performance errorMeasure, Node... inputs)
Sets up performance criteria and links.

Method Summary

protected double[]	coeffsFromData(double[] regressTo, double[][] trainingData) Least squares fitting of basic GMDH building-block, a (second-order) polynomial P(trainingData[0],trainingData[1]).
double[]	localOuputOnArray(double[][] inputs, double[] coefficients) Calculates the GMDH model output given an array of immediate inputs
protected double	localOutput(double[] inputs, double[] coefficients) Calculates this polynomial value given an array of immediate inputs
double	networkOutput(double[] inputs, int fold) Calculate total output of this GMDH model, recursively, given an instance.
double	networkOutputNoFlags(double[] inputs, int fold) Invoked by networkOutput.

Methods inherited from class wGmdh.jGmdh.oldskul.Model

coeffsAndErrors, coeffsAndErrors, coeffsAndErrors, collectTrainingData, collectValidationData, generateSummands, getCoeffs, getErrorMeasure, getErrorStructureLearningSet, getErrorStructureValidationSet, getSelectionCriterion, getStructureLearningGoals, getStructureLearningGoals, getStructureLearningInstances, getStructureLearningInstances, getStructureValidationGoals, getStructureValidationGoals, getStructureValidationInstances, getStructureValidationInstances, noLinks, setErrorMeasure, setSelectionCriterion

Methods inherited from class wGmdh.jGmdh.oldskul.Node

compareTo, getGlobalIdentifier, getIdentifier, getNrFolds, InitializeGraphics, setGlobalIdentifier

Methods inherited from class java.lang.Object

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

Field Detail

visitinfo

public TwoInputModel.VisitedHt visitinfo

Constructor Detail

TwoInputModel

public TwoInputModel(Node... links)

Parameters:

links -

TwoInputModel

public TwoInputModel(java.util.ArrayList<double[]> regressionGoal,
                     Performance selectionCriterion,
                     Performance errorMeasure,
                     Node inputL,
                     Node inputR)
              throws TooBig,
                     ExpressionEqualToZero,
                     TooSmall

Parameters:

regressionGoal -

selectionCriterion -

errorMeasure -

inputL -

inputR -

Throws:

jGMDH.exceptions.TooBig

jGMDH.exceptions.ExpressionEqualToZero

jGMDH.exceptions.TooSmall

TooBig

ExpressionEqualToZero

TooSmall

TwoInputModel

public TwoInputModel(double[] regressionGoals,
                     Performance selectionCriterion,
                     Performance errorMeasure,
                     Node... links)
              throws TooBig,
                     ExpressionEqualToZero,
                     TooSmall

Parameters:

regressionGoals -

selectionCriterion -

errorMeasure -

links -

Throws:

jGMDH.exceptions.TooBig

jGMDH.exceptions.ExpressionEqualToZero

jGMDH.exceptions.TooSmall

TooBig

ExpressionEqualToZero

TooSmall

TwoInputModel

public TwoInputModel(Performance selectionCriterion,
                     Performance errorMeasure,
                     Node... inputs)
              throws TooBig,
                     ExpressionEqualToZero,
                     TooSmall

Sets up performance criteria and links. Does not determine coefficients and measure quality of fit.

Parameters:

selectionCriterion -

errorMeasure -

inputs -

Throws:

TooBig

ExpressionEqualToZero

TooSmall

Method Detail

coeffsFromData

protected double[] coeffsFromData(double[] regressTo,
                                  double[][] trainingData)
                           throws TooBig

Least squares fitting of basic GMDH building-block, a (second-order) polynomial P(trainingData[0],trainingData[1]). Coefficients of the polynomial are treated as unknowns.

Specified by:

coeffsFromData in class Model

Parameters:

regressTo - values we are fitting to

trainingData -

Returns:

coefficients of polynomial

Throws:

TooBig

localOuputOnArray

public double[] localOuputOnArray(double[][] inputs,
                                  double[] coefficients)
                           throws TooBig

Calculates the GMDH model output given an array of immediate inputs

Specified by:

localOuputOnArray in class Model

Parameters:

inputs - array of length of polynomial inputs (number of inputs is 2)

coefficients -

Returns:

coefficients[0] + coefficients[1] * inputs[0][i] + coefficients[2] * inputs[1][i] + coefficients[3] * inputs[0][i] * inputs[0][i] + coefficients[4] * inputs[0][i] * inputs[1][i] + coefficients[5] * inputs[1][i] * inputs[1][i];, where i ranges from 0 to trainingData[0].length-1 == x2.length-1

Throws:

TooBig

localOutput

protected double localOutput(double[] inputs,
                             double[] coefficients)

Calculates this polynomial value given an array of immediate inputs

Parameters:

inputs - array of two doubles, one for each input

coefficients -

Returns:

coefficients[0] + coefficients[1] * inputs[0] + coefficients[2] * inputs[1] + coefficients[3] * inputs[0] * inputs[0] + coefficients[4] * inputs[0] * inputs[1] + coefficients[5] * inputs[1] * inputs[1];

networkOutput

public double networkOutput(double[] inputs,
                            int fold)

Calculate total output of this GMDH model, recursively, given an instance. Note: Models that share the same layer can be have their outputs calculated in parallel. Procesing of an array of inputs can also be optimized. (Athough, Weka doesn't have a classifyInstance(Instance[]), only classifyInstance(Instance)).

Specified by:

networkOutput in class Node

Parameters:

inputs - instance data organized as an array

fold - index of fold. if there are no multiple folds, pass 0

Returns:

networkOutputNoFlags

public double networkOutputNoFlags(double[] inputs,
                                   int fold)

Invoked by networkOutput.

Specified by:

networkOutputNoFlags in class Node

Parameters:

inputs -

fold -

Returns: