smile.clustering

Class NeuralMap

java.lang.Object

smile.clustering.NeuralMap

All Implemented Interfaces:

Clustering<double[]>

public class NeuralMap
extends Object
implements Clustering<double[]>

NeuralMap is an efficient competitive learning algorithm inspired by growing neural gas and BIRCH. Like growing neural gas, NeuralMap has the ability to add and delete neurons with competitive Hebbian learning. Edges exist between neurons close to each other. Such edges are intended place holders for localized data distribution. Such edges also help to locate distinct clusters (those clusters are not connected by edges). NeuralMap employs Locality-Sensitive Hashing to speedup the learning while BIRCH uses balanced CF trees.

Author:

Haifeng Li

See Also:

NeuralGas, GrowingNeuralGas, BIRCH

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	NeuralMap.Neuron The neurons in the network.

Field Summary

Fields inherited from interface smile.clustering.Clustering

OUTLIER

Constructor Summary

Constructors

Constructor and Description
NeuralMap(int d, double r, double epsBest, double epsNeighbor, int L, int k) Constructor.

Method Summary

Modifier and Type	Method and Description
List<NeuralMap.Neuron>	neurons() Returns the set of neurons.
void	partition(int k) Clustering neurons into k clusters.
int	partition(int k, int minPts) Clustering neurons into k clusters.
int	predict(double[] x) Cluster a new instance to the nearest neuron.
int	purge(int minPts) Removes neurons with the number of samples less than a given threshold.
void	update(double[] x) Update the network with a new signal.

Methods inherited from class java.lang.Object

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

Constructor Detail

NeuralMap

public NeuralMap(int d,
                 double r,
                 double epsBest,
                 double epsNeighbor,
                 int L,
                 int k)

Constructor.

Parameters:

d - the dimensionality of signals.

r - the distance radius to activate a neuron for a given signal.

epsBest - the fraction to update activated neuron.

epsNeighbor - the fraction to update neighbors of activated neuron.

L - the number of hash tables.

k - the number of random projection hash functions.

Method Detail

update

public void update(double[] x)

Update the network with a new signal.

neurons

public List<NeuralMap.Neuron> neurons()

Returns the set of neurons.

purge

public int purge(int minPts)

Removes neurons with the number of samples less than a given threshold. The neurons without neighbors will also be removed.

Parameters:

minPts - neurons will be removed if the number of its points is less than minPts.

Returns:

the number of neurons after purging.

partition

public void partition(int k)

Clustering neurons into k clusters.

Parameters:

k - the number of clusters.

partition

public int partition(int k,
                     int minPts)

Clustering neurons into k clusters.

Parameters:

k - the number of clusters.

minPts - a neuron will be treated as outlier if the number of its points is less than minPts.

Returns:

the number of non-outlier leaves.

predict

public int predict(double[] x)

Cluster a new instance to the nearest neuron. The method partition() should be called first.

Specified by:

predict in interface Clustering<double[]>

Parameters:

x - a new instance.

Returns:

the cluster label of nearest neuron.