Package smile.vq

Class NeuralMap

java.lang.Object

smile.vq.NeuralMap

All Implemented Interfaces:

Serializable, VectorQuantizer

public class NeuralMap
extends Object
implements VectorQuantizer

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 placeholders for localized data distribution. Such edges also help to locate distinct clusters (those clusters are not connected by edges).

See Also:

• NeuralGas
• GrowingNeuralGas
• BIRCH
• Serialized Form

Field Summary

Fields inherited from interface smile.vq.VectorQuantizer

OUTLIER

Constructor Summary

Constructors

Constructor	Description
NeuralMap(double r, double epsBest, double epsNeighbor, int edgeLifetime, double beta)	Constructor.

Method Summary

Modifier and Type	Method	Description
void	clear(double eps)	Removes staled neurons and the edges beyond lifetime.
Neuron[]	neurons()	Returns the neurons.
double[]	quantize(double[] x)	Quantize a new observation.
void	update(double[] x)	Update the codebook with a new observation.

Methods inherited from class java.lang.Object

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

Constructor Details

NeuralMap

public NeuralMap(double r,
 double epsBest,
 double epsNeighbor,
 int edgeLifetime,
 double beta)

Constructor.

Parameters:

r - the distance threshold to activate the nearest neuron of a signal.

epsBest - the learning rate to update activated neuron.

epsNeighbor - the learning rate to update neighbors of activated neuron.

edgeLifetime - the maximum age of edges.

beta - decrease the freshness of all neurons by multiply them with beta.

Method Details

update

public void update(double[] x)

Description copied from interface: VectorQuantizer

Update the codebook with a new observation.

Specified by:

update in interface VectorQuantizer

Parameters:

x - a new observation.

neurons

public Neuron[] neurons()

Returns the neurons.

Returns:

the neurons.

clear

public void clear(double eps)

Removes staled neurons and the edges beyond lifetime. Neurons without emanating edges will be removed too.

Parameters:

eps - the freshness threshold of neurons. It should be a small value (e.g. 1E-7).

quantize

public double[] quantize(double[] x)

Description copied from interface: VectorQuantizer

Quantize a new observation. Returns Optional.empty if the observation is noise.

Specified by:

quantize in interface VectorQuantizer

Parameters:

x - a new observation.

Returns:

the quantized vector.