jsat.datatransform

Class FastICA

java.lang.Object

jsat.datatransform.FastICA

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, DataTransform, InvertibleTransform

public class FastICA
extends java.lang.Object
implements InvertibleTransform

Provides an implementation of the FastICA algorithm for Independent Component Analysis (ICA). ICA is similar to PCA and Whitening, but assumes that the data is generated from a mixture of some C base components where mixing occurs instantaneously (i.e. produced from some matrix transform of the true components). ICA attempts to find the C components from the raw observations.

See:

• Hyvärinen, A. (1999). Fast and robust fixed-point algorithms for independent component analysis. IEEE Transactions on Neural Networks / a Publication of the IEEE Neural Networks Council, 10(3), 626–34. doi:10.1109/72.761722
• Hyvärinen, a.,&Oja, E. (2000). Independent component analysis: algorithms and applications. Neural Networks, 13(4-5), 411–430. doi:10.1016/S0893-6080(00)00026-5

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	FastICA.DefaultNegEntropyFunc A set of default negative entropy functions as specified in the original FastICA paper
static interface	FastICA.NegEntropyFunc The FastICA algorithm requires a function f(x) to be used iteratively in the algorithm, but only makes use of the first and second derivatives of the algorithm.

Constructor Summary

Constructors

Constructor and Description
FastICA() Creates a new FastICA transform that will attempt to fit 10 components.
FastICA(DataSet data, int C) Creates a new FastICA transform
FastICA(DataSet data, int C, FastICA.NegEntropyFunc G, boolean preWhitened) Creates a new FastICA transform
FastICA(FastICA toCopy) Copy constructor
FastICA(int C) Creates a new FastICA transform
FastICA(int C, FastICA.NegEntropyFunc G, boolean preWhitened) Creates a new FastICA transform

Method Summary

Modifier and Type	Method and Description
FastICA	clone()
void	fit(DataSet data) Fits this transform to the given dataset.
int	getC()
FastICA.NegEntropyFunc	getNegEntropyFunction()
DataPoint	inverse(DataPoint dp) Applies the inverse or "reverse" transform to approximately undo the effect of DataTransform.transform(jsat.classifiers.DataPoint) to recover an approximation of the original data point.
boolean	isPreWhitened()
void	setC(int C) Sets the number of base components to learn
void	setNegEntropyFunction(FastICA.NegEntropyFunc G) Sets the Negative Entropy function used to infer the base components.
void	setPreWhitened(boolean preWhitened) Controls where or not the implementation assumes the input data is already whitened.
DataPoint	transform(DataPoint dp) Returns a new data point that is a transformation of the original data point.

Methods inherited from class java.lang.Object

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

Constructor Detail

FastICA

public FastICA()

Creates a new FastICA transform that will attempt to fit 10 components. This is likely not optimal for any particular dataset

FastICA

public FastICA(int C)

Creates a new FastICA transform

Parameters:

C - the number of base components to assume and try to discover

FastICA

public FastICA(DataSet data,
               int C)

Creates a new FastICA transform

Parameters:

data - the data set to transform

C - the number of base components to assume and try to discover

FastICA

public FastICA(int C,
               FastICA.NegEntropyFunc G,
               boolean preWhitened)

Creates a new FastICA transform

Parameters:

data - the data set to transform

C - the number of base components to assume and try to discover

G - the Negative Entropy function to use

preWhitened - true to assume the data has already been whitened before being given to the transform, false and the FastICA implementation will perform its own whitening.

FastICA

public FastICA(DataSet data,
               int C,
               FastICA.NegEntropyFunc G,
               boolean preWhitened)

Creates a new FastICA transform

Parameters:

data - the data set to transform

C - the number of base components to assume and try to discover

G - the Negative Entropy function to use

preWhitened - true to assume the data has already been whitened before being given to the transform, false and the FastICA implementation will perform its own whitening.

FastICA

public FastICA(FastICA toCopy)

Copy constructor

Parameters:

toCopy - the object to copy

Method Detail

fit

public void fit(DataSet data)

Description copied from interface: DataTransform

Fits this transform to the given dataset. Some transforms can only be learned from classification or regression datasets. If an incompatible dataset type is given, a FailedToFitException exception may be thrown.

Specified by:

fit in interface DataTransform

Parameters:

data - the dataset to fir this transform to

setC

public void setC(int C)

Sets the number of base components to learn

Parameters:

C - the number of base components to assume and try to discover

getC

public int getC()

Returns:

the number of base components to assume and try to discover

setNegEntropyFunction

public void setNegEntropyFunction(FastICA.NegEntropyFunc G)

Sets the Negative Entropy function used to infer the base components.

Parameters:

G - the Negative Entropy function to use

getNegEntropyFunction

public FastICA.NegEntropyFunc getNegEntropyFunction()

Returns:

the Negative Entropy function to use

setPreWhitened

public void setPreWhitened(boolean preWhitened)

Controls where or not the implementation assumes the input data is already whitened. Whitening is a requirement for the algorithm to work as intended.

Parameters:

preWhitened - true to assume the data has already been whitened, false for this object to do its own whitening

isPreWhitened

public boolean isPreWhitened()

Returns:

true if this object will assume the data has already been whitened, false for this object to do its own whitening

transform

public DataPoint transform(DataPoint dp)

Description copied from interface: DataTransform

Returns a new data point that is a transformation of the original data point. This new data point is a different object, but may contain the same references as the original data point. It is not guaranteed that you can mutate the transformed point without having a side effect on the original point.

Specified by:

transform in interface DataTransform

Parameters:

dp - the data point to apply a transformation to

Returns:

a transformed data point

inverse

public DataPoint inverse(DataPoint dp)

Description copied from interface: InvertibleTransform

Applies the inverse or "reverse" transform to approximately undo the effect of DataTransform.transform(jsat.classifiers.DataPoint) to recover an approximation of the original data point.

Specified by:

inverse in interface InvertibleTransform

Parameters:

dp - the transformed data point

Returns:

the original data point, or a reasonable approximation

clone

public FastICA clone()

Specified by:

clone in interface DataTransform

Specified by:

clone in interface InvertibleTransform

Overrides:

clone in class java.lang.Object