gov.sandia.cognition.learning.algorithm.clustering

Class MiniBatchKMeansClusterer.Builder<DataType extends Vector>

java.lang.Object

gov.sandia.cognition.learning.algorithm.clustering.MiniBatchKMeansClusterer.Builder<DataType>

Type Parameters:

DataType -

Enclosing class:

MiniBatchKMeansClusterer<DataType extends Vector>

public static class MiniBatchKMeansClusterer.Builder<DataType extends Vector>
extends java.lang.Object

Can be used to create custom MiniBatchKMeansClusterers without using the big constructor.

Constructor Summary

Constructors

Constructor and Description
Builder(int numClusters) Create a mini-batch k-means clusterer builder and set it to the given number of clusters.
Builder(int numClusters, Semimetric<? super Vector> metric) Create a mini-batch k-means clusterer builder and set it to the given number of clusters.

Method Summary

Modifier and Type	Method and Description
MiniBatchKMeansClusterer<DataType>	build() Builds the clusterer.
MiniBatchKMeansClusterer.Builder<DataType>	withCreator(ClusterCreator<MiniBatchCentroidCluster,Vector> creator)
MiniBatchKMeansClusterer.Builder<DataType>	withInitializer(FixedClusterInitializer<MiniBatchCentroidCluster,Vector> initializer)
MiniBatchKMeansClusterer.Builder<DataType>	withMaxIterations(int maxIterations)
MiniBatchKMeansClusterer.Builder<DataType>	withMinibatchSize(int minibatchSize)
MiniBatchKMeansClusterer.Builder<DataType>	withNumClusters(int numClusters)
MiniBatchKMeansClusterer.Builder<DataType>	withRandom(java.util.Random random)

Methods inherited from class java.lang.Object

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

Constructor Detail

Builder

public Builder(int numClusters)

Create a mini-batch k-means clusterer builder and set it to the given number of clusters. Centroids will be initialized using a GreedyClusterInitializer.

Parameters:

numClusters -

Builder

public Builder(int numClusters,
               Semimetric<? super Vector> metric)

Create a mini-batch k-means clusterer builder and set it to the given number of clusters. Centroids will be initialized using a GreedyClusterInitializer, and the given metric will be used to measure all distances.

Parameters:

numClusters -

metric - the semimetric to use to measure distances

Method Detail

build

public MiniBatchKMeansClusterer<DataType> build()

Builds the clusterer.

Returns:

The newly built clusterer.

withNumClusters

public MiniBatchKMeansClusterer.Builder<DataType> withNumClusters(int numClusters)

Parameters:

numClusters - the number of clusters to create

Returns:

the builder

withMaxIterations

public MiniBatchKMeansClusterer.Builder<DataType> withMaxIterations(int maxIterations)

Parameters:

maxIterations - the number of iterations before stopping

Returns:

the builder

withMinibatchSize

public MiniBatchKMeansClusterer.Builder<DataType> withMinibatchSize(int minibatchSize)

Parameters:

minibatchSize - the mini-batch size

Returns:

the builder

See Also:

MiniBatchKMeansClusterer.setMinibatchSize(int)

withInitializer

public MiniBatchKMeansClusterer.Builder<DataType> withInitializer(FixedClusterInitializer<MiniBatchCentroidCluster,Vector> initializer)

Parameters:

initializer - sets the initial centroids

Returns:

the builder

withCreator

public MiniBatchKMeansClusterer.Builder<DataType> withCreator(ClusterCreator<MiniBatchCentroidCluster,Vector> creator)

Parameters:

creator - the cluster creator to use

Returns:

the builder

withRandom

public MiniBatchKMeansClusterer.Builder<DataType> withRandom(java.util.Random random)

Parameters:

random - the random number generator to use

Returns:

the builder