org.deeplearning4j.optimize

Class MultiLayerNetworkOptimizer

java.lang.Object

org.deeplearning4j.optimize.MultiLayerNetworkOptimizer

All Implemented Interfaces:

cc.mallet.optimize.Optimizable, cc.mallet.optimize.Optimizable.ByGradientValue, Serializable, OptimizableByGradientValueMatrix

public class MultiLayerNetworkOptimizer
extends Object
implements cc.mallet.optimize.Optimizable.ByGradientValue, Serializable, OptimizableByGradientValueMatrix

Optimizes the logistic layer for finetuning a multi layer network. This is meant to be used after pretraining.

Author:

Adam Gibson

See Also:

Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from interface cc.mallet.optimize.Optimizable

cc.mallet.optimize.Optimizable.ByBatchGradient, cc.mallet.optimize.Optimizable.ByCombiningBatchGradient, cc.mallet.optimize.Optimizable.ByGISUpdate, cc.mallet.optimize.Optimizable.ByGradient, cc.mallet.optimize.Optimizable.ByGradientValue, cc.mallet.optimize.Optimizable.ByHessian, cc.mallet.optimize.Optimizable.ByValue, cc.mallet.optimize.Optimizable.ByVotedPerceptron

Field Summary

Fields

Modifier and Type	Field and Description
protected BaseMultiLayerNetwork	network

Constructor Summary

Constructors

Constructor and Description
MultiLayerNetworkOptimizer(BaseMultiLayerNetwork network, double lr)

Method Summary

Methods

Modifier and Type	Method and Description
int	getNumParameters()
double	getParameter(int index)
org.jblas.DoubleMatrix	getParameters()
void	getParameters(double[] buffer)
double	getValue()
org.jblas.DoubleMatrix	getValueGradient()
void	getValueGradient(double[] buffer)
void	optimize(org.jblas.DoubleMatrix labels, double lr, int epochs)
void	setParameter(int index, double value)
void	setParameters(double[] params)
void	setParameters(org.jblas.DoubleMatrix params)

Methods inherited from class java.lang.Object

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

Field Detail

network

protected BaseMultiLayerNetwork network

Constructor Detail

MultiLayerNetworkOptimizer

public MultiLayerNetworkOptimizer(BaseMultiLayerNetwork network,
                          double lr)

Method Detail

optimize

public void optimize(org.jblas.DoubleMatrix labels,
            double lr,
            int epochs)

getNumParameters

public int getNumParameters()

Specified by:

getNumParameters in interface cc.mallet.optimize.Optimizable

Specified by:

getNumParameters in interface OptimizableByGradientValueMatrix

getParameters

public void getParameters(double[] buffer)

Specified by:

getParameters in interface cc.mallet.optimize.Optimizable

getParameter

public double getParameter(int index)

Specified by:

getParameter in interface cc.mallet.optimize.Optimizable

Specified by:

getParameter in interface OptimizableByGradientValueMatrix

setParameters

public void setParameters(double[] params)

Specified by:

setParameters in interface cc.mallet.optimize.Optimizable

setParameter

public void setParameter(int index,
                double value)

Specified by:

setParameter in interface cc.mallet.optimize.Optimizable

Specified by:

setParameter in interface OptimizableByGradientValueMatrix

getValueGradient

public void getValueGradient(double[] buffer)

Specified by:

getValueGradient in interface cc.mallet.optimize.Optimizable.ByGradientValue

getValue

public double getValue()

Specified by:

getValue in interface cc.mallet.optimize.Optimizable.ByGradientValue

Specified by:

getValue in interface OptimizableByGradientValueMatrix

getParameters

public org.jblas.DoubleMatrix getParameters()

Specified by:

getParameters in interface OptimizableByGradientValueMatrix

setParameters

public void setParameters(org.jblas.DoubleMatrix params)

Specified by:

setParameters in interface OptimizableByGradientValueMatrix

getValueGradient

public org.jblas.DoubleMatrix getValueGradient()

Specified by:

getValueGradient in interface OptimizableByGradientValueMatrix