org.deeplearning4j.nn

Interface NeuralNetwork

All Superinterfaces:

Cloneable, NeuralNetEpochListener, Serializable

All Known Implementing Classes:

BaseNeuralNetwork, ConvolutionalRBM, CRBM, DenoisingAutoEncoder, GaussianRectifiedLinearRBM, RBM

public interface NeuralNetwork
extends Serializable, Cloneable, NeuralNetEpochListener

Single layer neural network, this is typically one that has the objective function of reconstruction the input: also called feature detectors

Author:

Adam Gibson

Nested Class Summary

Nested Classes

Modifier and Type	Interface and Description
static class	NeuralNetwork.LossFunction Which loss function to use
static class	NeuralNetwork.OptimizationAlgorithm Optimization algorithm to use

Method Summary

Methods

Modifier and Type	Method and Description
NeuralNetwork	clone()
double	dropOut()
void	epochDone(int epoch) Event listener for each iteration
double	fanIn()
AdaGrad	getAdaGrad()
org.apache.commons.math3.distribution.RealDistribution	getDist()
NeuralNetworkGradient	getGradient(Object[] params)
List<NeuralNetworkGradientListener>	getGradientListeners()
org.jblas.DoubleMatrix	gethBias()
AdaGrad	gethBiasAdaGrad()
org.jblas.DoubleMatrix	getInput()
double	getL2()
NeuralNetwork.LossFunction	getLossFunction()
double	getMomentum()
int	getnHidden()
int	getnVisible()
NeuralNetwork.OptimizationAlgorithm	getOptimizationAlgorithm()
double	getReConstructionCrossEntropy() Error on reconstruction
int	getRenderEpochs()
org.apache.commons.math3.random.RandomGenerator	getRng()
double	getSparsity()
org.jblas.DoubleMatrix	getvBias()
AdaGrad	getVBiasAdaGrad()
org.jblas.DoubleMatrix	getW()
org.jblas.DoubleMatrix	hBiasMean()
double	l2RegularizedCoefficient()
void	merge(NeuralNetwork network, int batchSize) Performs a network merge in the form of a += b - a / n where a is a matrix here b is a matrix on the incoming network and n is the batch size
double	negativeLogLikelihood()
boolean	normalizeByInputRows()
void	resetAdaGrad(double lr)
Pair<org.jblas.DoubleMatrix,org.jblas.DoubleMatrix>	sampleHiddenGivenVisible(org.jblas.DoubleMatrix v) Sample hidden mean and sample given visible
Pair<org.jblas.DoubleMatrix,org.jblas.DoubleMatrix>	sampleVisibleGivenHidden(org.jblas.DoubleMatrix h) Sample visible mean and sample given hidden
void	setAdaGrad(AdaGrad adaGrad)
void	setDist(org.apache.commons.math3.distribution.RealDistribution dist)
void	setDropOut(double dropOut)
void	setFanIn(double fanIn)
void	setGradientListeners(List<NeuralNetworkGradientListener> gradientListeners)
void	sethBias(org.jblas.DoubleMatrix hBias)
void	setHbiasAdaGrad(AdaGrad adaGrad)
void	setInput(org.jblas.DoubleMatrix input)
void	setL2(double l2)
void	setLossFunction(NeuralNetwork.LossFunction lossFunction)
void	setMomentum(double momentum)
void	setnHidden(int nHidden)
void	setnVisible(int nVisible)
void	setOptimizationAlgorithm(NeuralNetwork.OptimizationAlgorithm optimziationAlgorithm)
void	setRenderEpochs(int renderEpochs)
void	setRng(org.apache.commons.math3.random.RandomGenerator rng)
void	setSparsity(double sparsity)
void	setvBias(org.jblas.DoubleMatrix vBias)
void	setVBiasAdaGrad(AdaGrad adaGrad)
void	setW(org.jblas.DoubleMatrix w)
double	squaredLoss()
void	train(org.jblas.DoubleMatrix input, double lr, Object[] params) Run one iteration
void	trainTillConvergence(org.jblas.DoubleMatrix input, double lr, Object[] params) Trains via an optimization algorithm such as SGD or Conjugate Gradient
NeuralNetwork	transpose()

Method Detail

getLossFunction

NeuralNetwork.LossFunction getLossFunction()

setLossFunction

void setLossFunction(NeuralNetwork.LossFunction lossFunction)

getOptimizationAlgorithm

NeuralNetwork.OptimizationAlgorithm getOptimizationAlgorithm()

setOptimizationAlgorithm

void setOptimizationAlgorithm(NeuralNetwork.OptimizationAlgorithm optimziationAlgorithm)

normalizeByInputRows

boolean normalizeByInputRows()

getnVisible

int getnVisible()

setnVisible

void setnVisible(int nVisible)

getnHidden

int getnHidden()

setnHidden

void setnHidden(int nHidden)

getW

org.jblas.DoubleMatrix getW()

setW

void setW(org.jblas.DoubleMatrix w)

gethBias

org.jblas.DoubleMatrix gethBias()

sethBias

void sethBias(org.jblas.DoubleMatrix hBias)

getvBias

org.jblas.DoubleMatrix getvBias()

setvBias

void setvBias(org.jblas.DoubleMatrix vBias)

getRng

org.apache.commons.math3.random.RandomGenerator getRng()

setRng

void setRng(org.apache.commons.math3.random.RandomGenerator rng)

getInput

org.jblas.DoubleMatrix getInput()

setInput

void setInput(org.jblas.DoubleMatrix input)

squaredLoss

double squaredLoss()

negativeLogLikelihood

double negativeLogLikelihood()

getSparsity

double getSparsity()

setSparsity

void setSparsity(double sparsity)

setDist

void setDist(org.apache.commons.math3.distribution.RealDistribution dist)

getDist

org.apache.commons.math3.distribution.RealDistribution getDist()

getGradientListeners

List<NeuralNetworkGradientListener> getGradientListeners()

setGradientListeners

void setGradientListeners(List<NeuralNetworkGradientListener> gradientListeners)

hBiasMean

org.jblas.DoubleMatrix hBiasMean()

getAdaGrad

AdaGrad getAdaGrad()

setAdaGrad

void setAdaGrad(AdaGrad adaGrad)

gethBiasAdaGrad

AdaGrad gethBiasAdaGrad()

setHbiasAdaGrad

void setHbiasAdaGrad(AdaGrad adaGrad)

getVBiasAdaGrad

AdaGrad getVBiasAdaGrad()

setVBiasAdaGrad

void setVBiasAdaGrad(AdaGrad adaGrad)

getGradient

NeuralNetworkGradient getGradient(Object[] params)

getL2

double getL2()

setL2

void setL2(double l2)

getMomentum

double getMomentum()

setMomentum

void setMomentum(double momentum)

setRenderEpochs

void setRenderEpochs(int renderEpochs)

getRenderEpochs

int getRenderEpochs()

transpose

NeuralNetwork transpose()

clone

NeuralNetwork clone()

fanIn

double fanIn()

setFanIn

void setFanIn(double fanIn)

sampleHiddenGivenVisible

Pair<org.jblas.DoubleMatrix,org.jblas.DoubleMatrix> sampleHiddenGivenVisible(org.jblas.DoubleMatrix v)

Sample hidden mean and sample given visible

Parameters:

v - the the visible input

Returns:

a pair with mean, sample

setDropOut

void setDropOut(double dropOut)

dropOut

double dropOut()

sampleVisibleGivenHidden

Pair<org.jblas.DoubleMatrix,org.jblas.DoubleMatrix> sampleVisibleGivenHidden(org.jblas.DoubleMatrix h)

Sample visible mean and sample given hidden

Parameters:

h - the the hidden input

Returns:

a pair with mean, sample

resetAdaGrad

void resetAdaGrad(double lr)

epochDone

void epochDone(int epoch)

Description copied from interface: NeuralNetEpochListener

Event listener for each iteration

Specified by:

epochDone in interface NeuralNetEpochListener

l2RegularizedCoefficient

double l2RegularizedCoefficient()

getReConstructionCrossEntropy

double getReConstructionCrossEntropy()

Error on reconstruction

Returns:

the error on reconstruction

train

void train(org.jblas.DoubleMatrix input,
         double lr,
         Object[] params)

Run one iteration

Parameters:

input - the input to train on

lr - the learning rate to use

params - the extra params for the neural network(k, corruption level, max epochs,...)

trainTillConvergence

void trainTillConvergence(org.jblas.DoubleMatrix input,
                        double lr,
                        Object[] params)

Trains via an optimization algorithm such as SGD or Conjugate Gradient

Parameters:

input - the input to train on

lr - the learning rate to use

params - the params (k,corruption level, max epochs,...)

merge

void merge(NeuralNetwork network,
         int batchSize)

Performs a network merge in the form of a += b - a / n where a is a matrix here b is a matrix on the incoming network and n is the batch size

Parameters:

network - the network to merge with

batchSize - the batch size (number of training examples) to average by