org.deeplearning4j.plot

Class BarnesHutTsne

java.lang.Object

org.deeplearning4j.plot.BarnesHutTsne

All Implemented Interfaces:

Model

public class BarnesHutTsne
extends Object
implements Model

Barnes hut algorithm for TSNE, uses a dual tree approximation approach. Work based on: http://lvdmaaten.github.io/tsne/ For hight dimensions, it's recommended to reduce the dimension up to 50 using another method (PCA or other)

Author:

Adam Gibson

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	BarnesHutTsne.Builder

Field Summary

Fields

Modifier and Type	Field and Description
protected org.nd4j.linalg.learning.legacy.AdaGrad	adaGrad
protected double	finalMomentum
protected double	initialMomentum
protected IterationListener	iterationListener
protected double	learningRate
protected int	maxIter
protected double	minGain
protected double	momentum
protected boolean	normalize
protected double	perplexity
protected double	realMin
protected int	stopLyingIteration
protected int	switchMomentumIteration
protected double	tolerance
protected boolean	useAdaGrad
protected boolean	usePca
protected org.nd4j.linalg.api.ndarray.INDArray	Y
static String	Y_GRAD

Constructor Summary

Constructors

Constructor and Description
BarnesHutTsne(int numDimensions, String simiarlityFunction, double theta, boolean invert, int maxIter, double realMin, double initialMomentum, double finalMomentum, double momentum, int switchMomentumIteration, boolean normalize, int stopLyingIteration, double tolerance, double learningRate, boolean useAdaGrad, double perplexity, IterationListener iterationListener, double minGain)

Method Summary

Modifier and Type	Method and Description
void	accumulateScore(double accum)
void	addListeners(IterationListener... listener)
void	applyLearningRateScoreDecay()
int	batchSize()
void	clear()
org.nd4j.linalg.primitives.Pair<org.nd4j.linalg.api.ndarray.INDArray,Double>	computeGaussianKernel(org.nd4j.linalg.api.ndarray.INDArray distances, double beta, int k) Computes a gaussian kernel given a vector of squared distance distances
org.nd4j.linalg.api.ndarray.INDArray	computeGaussianPerplexity(org.nd4j.linalg.api.ndarray.INDArray d, double u) Convert data to probability co-occurrences (aka calculating the kernel)
void	computeGradientAndScore()
NeuralNetConfiguration	conf()
void	fit()
void	fit(org.nd4j.linalg.api.ndarray.INDArray data)
void	fit(org.nd4j.linalg.api.ndarray.INDArray data, int nDims) Deprecated. Use fit(INDArray)
org.nd4j.linalg.api.ndarray.INDArray	getData() Return the matrix reduce to the NDim.
org.nd4j.linalg.api.ndarray.INDArray	getGradientsViewArray()
int	getNumDimensions()
ConvexOptimizer	getOptimizer()
org.nd4j.linalg.api.ndarray.INDArray	getParam(String param)
double	getPerplexity()
String	getSimiarlityFunction()
double	getTheta()
Gradient	gradient()
protected org.nd4j.linalg.primitives.Pair<Double,org.nd4j.linalg.api.ndarray.INDArray>	gradient(org.nd4j.linalg.api.ndarray.INDArray p)
org.nd4j.linalg.primitives.Pair<Gradient,Double>	gradientAndScore()
void	init() Init the model
void	initParams()
org.nd4j.linalg.api.ndarray.INDArray	input()
boolean	isInvert()
void	iterate(org.nd4j.linalg.api.ndarray.INDArray input)
int	numParams()
int	numParams(boolean backwards)
org.nd4j.linalg.api.ndarray.INDArray	params()
Map<String,org.nd4j.linalg.api.ndarray.INDArray>	paramTable()
Map<String,org.nd4j.linalg.api.ndarray.INDArray>	paramTable(boolean backprapParamsOnly)
void	plot(org.nd4j.linalg.api.ndarray.INDArray matrix, int nDims, List<String> labels, String path) Deprecated. use fit(INDArray) and saveAsFile(List, String) instead.
void	saveAsFile(List<String> labels, String path) Save the model as a file with a csv format, adding the label as the last column.
double	score()
void	setBackpropGradientsViewArray(org.nd4j.linalg.api.ndarray.INDArray gradients)
void	setConf(NeuralNetConfiguration conf)
void	setData(org.nd4j.linalg.api.ndarray.INDArray data)
void	setInvert(boolean invert)
void	setListeners(Collection<IterationListener> listeners) Set the IterationListeners for the ComputationGraph (and all layers in the network)
void	setListeners(IterationListener... listeners) Set the IterationListeners for the ComputationGraph (and all layers in the network)
void	setNumDimensions(int numDimensions)
void	setParam(String key, org.nd4j.linalg.api.ndarray.INDArray val)
void	setParams(org.nd4j.linalg.api.ndarray.INDArray params)
void	setParamsViewArray(org.nd4j.linalg.api.ndarray.INDArray params)
void	setParamTable(Map<String,org.nd4j.linalg.api.ndarray.INDArray> paramTable)
void	setSimiarlityFunction(String simiarlityFunction)
void	step(org.nd4j.linalg.api.ndarray.INDArray p, int i) An individual iteration
org.nd4j.linalg.api.ndarray.INDArray	symmetrized(org.nd4j.linalg.api.ndarray.INDArray rowP, org.nd4j.linalg.api.ndarray.INDArray colP, org.nd4j.linalg.api.ndarray.INDArray valP) Symmetrize the value matrix
void	update(Gradient gradient)
void	update(org.nd4j.linalg.api.ndarray.INDArray gradient, String paramType)
void	validateInput()

Methods inherited from class java.lang.Object

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

Field Detail

maxIter

protected int maxIter

realMin

protected double realMin

initialMomentum

protected double initialMomentum

finalMomentum

protected double finalMomentum

minGain

protected double minGain

momentum

protected double momentum

switchMomentumIteration

protected int switchMomentumIteration

normalize

protected boolean normalize

usePca

protected boolean usePca

stopLyingIteration

protected int stopLyingIteration

tolerance

protected double tolerance

learningRate

protected double learningRate

adaGrad

protected org.nd4j.linalg.learning.legacy.AdaGrad adaGrad

useAdaGrad

protected boolean useAdaGrad

perplexity

protected double perplexity

Y

protected org.nd4j.linalg.api.ndarray.INDArray Y

Y_GRAD

public static final String Y_GRAD

See Also:

Constant Field Values

iterationListener

protected transient IterationListener iterationListener

Constructor Detail

BarnesHutTsne

public BarnesHutTsne(int numDimensions,
                     String simiarlityFunction,
                     double theta,
                     boolean invert,
                     int maxIter,
                     double realMin,
                     double initialMomentum,
                     double finalMomentum,
                     double momentum,
                     int switchMomentumIteration,
                     boolean normalize,
                     int stopLyingIteration,
                     double tolerance,
                     double learningRate,
                     boolean useAdaGrad,
                     double perplexity,
                     IterationListener iterationListener,
                     double minGain)

Method Detail

getSimiarlityFunction

public String getSimiarlityFunction()

setSimiarlityFunction

public void setSimiarlityFunction(String simiarlityFunction)

isInvert

public boolean isInvert()

setInvert

public void setInvert(boolean invert)

getTheta

public double getTheta()

getPerplexity

public double getPerplexity()

getNumDimensions

public int getNumDimensions()

setNumDimensions

public void setNumDimensions(int numDimensions)

computeGaussianPerplexity

public org.nd4j.linalg.api.ndarray.INDArray computeGaussianPerplexity(org.nd4j.linalg.api.ndarray.INDArray d,
                                                                      double u)

Convert data to probability co-occurrences (aka calculating the kernel)

Parameters:

d - the data to convert

u - the perplexity of the model

Returns:

the probabilities of co-occurrence

input

public org.nd4j.linalg.api.ndarray.INDArray input()

Specified by:

input in interface Model

validateInput

public void validateInput()

Specified by:

validateInput in interface Model

getOptimizer

public ConvexOptimizer getOptimizer()

Specified by:

getOptimizer in interface Model

getParam

public org.nd4j.linalg.api.ndarray.INDArray getParam(String param)

Specified by:

getParam in interface Model

initParams

public void initParams()

Specified by:

initParams in interface Model

addListeners

public void addListeners(IterationListener... listener)

Specified by:

addListeners in interface Model

paramTable

public Map<String,org.nd4j.linalg.api.ndarray.INDArray> paramTable()

Specified by:

paramTable in interface Model

paramTable

public Map<String,org.nd4j.linalg.api.ndarray.INDArray> paramTable(boolean backprapParamsOnly)

Specified by:

paramTable in interface Model

setParamTable

public void setParamTable(Map<String,org.nd4j.linalg.api.ndarray.INDArray> paramTable)

Specified by:

setParamTable in interface Model

setParam

public void setParam(String key,
                     org.nd4j.linalg.api.ndarray.INDArray val)

Specified by:

setParam in interface Model

clear

public void clear()

Specified by:

clear in interface Model

gradient

protected org.nd4j.linalg.primitives.Pair<Double,org.nd4j.linalg.api.ndarray.INDArray> gradient(org.nd4j.linalg.api.ndarray.INDArray p)

symmetrized

public org.nd4j.linalg.api.ndarray.INDArray symmetrized(org.nd4j.linalg.api.ndarray.INDArray rowP,
                                                        org.nd4j.linalg.api.ndarray.INDArray colP,
                                                        org.nd4j.linalg.api.ndarray.INDArray valP)

Symmetrize the value matrix

Parameters:

rowP -

colP -

valP -

Returns:

computeGaussianKernel

public org.nd4j.linalg.primitives.Pair<org.nd4j.linalg.api.ndarray.INDArray,Double> computeGaussianKernel(org.nd4j.linalg.api.ndarray.INDArray distances,
                                                                                                          double beta,
                                                                                                          int k)

Computes a gaussian kernel given a vector of squared distance distances

Parameters:

distances -

beta -

Returns:

init

public void init()

Init the model

Specified by:

init in interface Model

setListeners

public void setListeners(Collection<IterationListener> listeners)

Set the IterationListeners for the ComputationGraph (and all layers in the network)

Specified by:

setListeners in interface Model

Parameters:

listeners -

setListeners

public void setListeners(IterationListener... listeners)

Set the IterationListeners for the ComputationGraph (and all layers in the network)

Specified by:

setListeners in interface Model

Parameters:

listeners -

fit

public void fit()

Specified by:

fit in interface Model

update

public void update(Gradient gradient)

Specified by:

update in interface Model

step

public void step(org.nd4j.linalg.api.ndarray.INDArray p,
                 int i)

An individual iteration

Parameters:

p - the probabilities that certain points are near each other

i - the iteration (primarily for debugging purposes)

update

public void update(org.nd4j.linalg.api.ndarray.INDArray gradient,
                   String paramType)

Specified by:

update in interface Model

saveAsFile

public void saveAsFile(List<String> labels,
                       String path)
                throws IOException

Save the model as a file with a csv format, adding the label as the last column.

Parameters:

labels -

path - the path to write

Throws:

IOException

plot

@Deprecated
public void plot(org.nd4j.linalg.api.ndarray.INDArray matrix,
                             int nDims,
                             List<String> labels,
                             String path)
                      throws IOException

Deprecated. use fit(INDArray) and saveAsFile(List, String) instead.

Plot tsne

Parameters:

matrix - the matrix to plot

nDims - the number

labels -

path - the path to write

Throws:

IOException

score

public double score()

Specified by:

score in interface Model

computeGradientAndScore

public void computeGradientAndScore()

Specified by:

computeGradientAndScore in interface Model

accumulateScore

public void accumulateScore(double accum)

Specified by:

accumulateScore in interface Model

params

public org.nd4j.linalg.api.ndarray.INDArray params()

Specified by:

params in interface Model

numParams

public int numParams()

Specified by:

numParams in interface Model

numParams

public int numParams(boolean backwards)

Specified by:

numParams in interface Model

setParams

public void setParams(org.nd4j.linalg.api.ndarray.INDArray params)

Specified by:

setParams in interface Model

setParamsViewArray

public void setParamsViewArray(org.nd4j.linalg.api.ndarray.INDArray params)

Specified by:

setParamsViewArray in interface Model

getGradientsViewArray

public org.nd4j.linalg.api.ndarray.INDArray getGradientsViewArray()

Specified by:

getGradientsViewArray in interface Model

setBackpropGradientsViewArray

public void setBackpropGradientsViewArray(org.nd4j.linalg.api.ndarray.INDArray gradients)

Specified by:

setBackpropGradientsViewArray in interface Model

applyLearningRateScoreDecay

public void applyLearningRateScoreDecay()

Specified by:

applyLearningRateScoreDecay in interface Model

fit

public void fit(org.nd4j.linalg.api.ndarray.INDArray data)

Specified by:

fit in interface Model

fit

@Deprecated
public void fit(org.nd4j.linalg.api.ndarray.INDArray data,
                            int nDims)

Deprecated. Use fit(INDArray)

Change the dimensions with

iterate

public void iterate(org.nd4j.linalg.api.ndarray.INDArray input)

Specified by:

iterate in interface Model

gradient

public Gradient gradient()

Specified by:

gradient in interface Model

gradientAndScore

public org.nd4j.linalg.primitives.Pair<Gradient,Double> gradientAndScore()

Specified by:

gradientAndScore in interface Model

batchSize

public int batchSize()

Specified by:

batchSize in interface Model

conf

public NeuralNetConfiguration conf()

Specified by:

conf in interface Model

setConf

public void setConf(NeuralNetConfiguration conf)

Specified by:

setConf in interface Model

getData

public org.nd4j.linalg.api.ndarray.INDArray getData()

Return the matrix reduce to the NDim.

setData

public void setData(org.nd4j.linalg.api.ndarray.INDArray data)