org.imirsel.m2k.modelling
Class Gaussian

java.lang.Object
  org.imirsel.m2k.modelling.Gaussian

All Implemented Interfaces:

java.io.Serializable

public class Gaussian

extends java.lang.Object

implements java.io.Serializable

A single Gaussian classifier with diagonal covariance

Author:

kw

See Also:

Serialized Form

Constructor Summary

Gaussian()
Creates a new instance of Gaussian

Gaussian(Gaussian oldGaussian)
Copy constructor for a Gaussian

Gaussian(int numClasses_)
Creates a new instance of Gaussian

Method Summary

int	classify(double[] inputVector) Calculates the mahalanobis distances of the inputVector from each class and returns the integer class label of the closest class
double	mahalanobis(double[] vec, double[] means, double[] covars) Calculates the mahalanobis distance of the input vector from the means and covariances of the Gaussian model supplied
void	train(double[][] inputData, int[] labels) Trains the Gaussian model using diagonal covariance

Methods inherited from class java.lang.Object

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

Constructor Detail

Gaussian

public Gaussian()

Creates a new instance of Gaussian

Gaussian

public Gaussian(int numClasses_)

Creates a new instance of Gaussian

Parameters:

numClasses_ - Num of classes classifier is to be trained on

Gaussian

public Gaussian(Gaussian oldGaussian)

Copy constructor for a Gaussian

Parameters:

oldGaussian - The Gaussian to copied

Method Detail

classify

public int classify(double[] inputVector)

Calculates the mahalanobis distances of the inputVector from each class and returns the integer class label of the closest class

Parameters:

inputVector - The vector to be classified

Returns:

the integer class label of the closest class

train

public void train(double[][] inputData,
                  int[] labels)

Trains the Gaussian model using diagonal covariance

Parameters:

inputData - The data to train the model on

labels - The class labels of the input data

mahalanobis

public double mahalanobis(double[] vec,
                          double[] means,
                          double[] covars)

Calculates the mahalanobis distance of the input vector from the means and covariances of the Gaussian model supplied

Parameters:

vec - The vector to calculate the mahalanobis distance for

means - The means of the Gaussian model to test against

covars - The covariances of the Gaussian model to test against

Returns:

The mahalanobis distance