marytts.machinelearning

Class PolynomialHierarchicalClusteringTrainer

java.lang.Object

marytts.machinelearning.PolynomialHierarchicalClusteringTrainer

public class PolynomialHierarchicalClusteringTrainer
extends Object

Hierarchical clustering training algorithm Reference: Stephen C. Johnson, 1967, "Hierarchical clustering schemes", Proc. Psychometrika, Vol. 32 No. 3, pp. 241-254. This version is adapted to work with a distance function between polynomials.

Author:

Sathish Pammi

Constructor Summary

Constructors

Constructor and Description
PolynomialHierarchicalClusteringTrainer(Polynomial[] polynomials) Constructor of the Hierarchical trainer

Method Summary

Methods

Modifier and Type	Method and Description
static void	main(String[] args) Main method
PolynomialCluster[]	train(int tagetClusterSize, String linkageType) This function clusters polynomials using Hierarchical (agglomerative approach) clustering procedure, using a polynomial distance function.

Methods inherited from class java.lang.Object

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

Constructor Detail

PolynomialHierarchicalClusteringTrainer

public PolynomialHierarchicalClusteringTrainer(Polynomial[] polynomials)

Constructor of the Hierarchical trainer

Parameters:

polynomials - - array of polynomial coefficients ( minimum length of polynomials should be three )

Throws:

NullPointerException - if input polynomial array is null

IllegalArgumentException - if length of array of polynomial coefficients is less than three

Method Detail

train

public PolynomialCluster[] train(int tagetClusterSize,
                        String linkageType)

This function clusters polynomials using Hierarchical (agglomerative approach) clustering procedure, using a polynomial distance function. Training consists of four steps: 1. Convert object features to distance matrix. 2. Set each object as a cluster (thus if we have 6 objects, we will have 6 clusters in the beginning) 3. Iterate until number of clusters is equal to the given target number of clusters - Merge two closest clusters - Update distance matrix

Parameters:

tagetClusterSize - the target cluster size

linkageType - the linkage type used for Hierarchical clustering ('Average', 'Complete', or 'Short')

Returns:

the trained clusters

Throws:

IllegalArgumentException - if target cluster size is not less than initialized number of samples

main

public static void main(String[] args)

Main method

Parameters:

args - args