de.lmu.ifi.dbs.elki.algorithm.clustering.em

Interface EMClusterModel<M extends MeanModel>

All Known Implementing Classes:

DiagonalGaussianModel, MultivariateGaussianModel, SphericalGaussianModel, TextbookMultivariateGaussianModel, TwoPassMultivariateGaussianModel

public interface EMClusterModel<M extends MeanModel>

Models useable in EM clustering.

Since:

0.7.0

Author:

Erich Schubert

Method Summary

Modifier and Type	Method and Description
void	beginEStep() Begin the E step.
double	estimateLogDensity(NumberVector vec) Estimate the log likelihood of a vector.
M	finalizeCluster() Finalize a cluster model.
void	finalizeEStep(double weight, double prior) Finalize the E step.
default void	finalizeFirstPassE() Finalize the first pass of the E step.
default void	firstPassE(NumberVector vec, double weight) First run in the E step.
double	getWeight() Get the cluster weight.
default boolean	needsTwoPass() True, if the model needs two passes in the E step.
void	setWeight(double weight) Set the cluster weight.
void	updateE(NumberVector vec, double weight) Process one data point in the E step

Method Detail

beginEStep

void beginEStep()

Begin the E step.

needsTwoPass

default boolean needsTwoPass()

True, if the model needs two passes in the E step.

Returns:

true when an initial pass is needed.

firstPassE

default void firstPassE(NumberVector vec,
                        double weight)

First run in the E step. By default, this is not used (c.f. needsTwoPass().

Parameters:

vec - Vector to process

weight - Weight of point ("responsibility" of the cluster)

finalizeFirstPassE

default void finalizeFirstPassE()

Finalize the first pass of the E step. By default, this is not used (c.f. needsTwoPass().

updateE

void updateE(NumberVector vec,
             double weight)

Process one data point in the E step

Parameters:

vec - Vector to process

weight - Weight of point ("responsibility" of the cluster)

finalizeEStep

void finalizeEStep(double weight,
                   double prior)

Finalize the E step.

Parameters:

weight - weight of the cluster

prior - MAP prior (0 for MLE)

estimateLogDensity

double estimateLogDensity(NumberVector vec)

Estimate the log likelihood of a vector.

Parameters:

vec - Vector

Returns:

log likelihood.

finalizeCluster

M finalizeCluster()

Finalize a cluster model.

Returns:

Cluster model

getWeight

double getWeight()

Get the cluster weight.

Returns:

Cluster weight

setWeight

void setWeight(double weight)

Set the cluster weight.

Parameters:

weight - Cluster weight