org.nd4j.linalg.learning

Class AdaGrad

java.lang.Object

org.nd4j.linalg.learning.AdaGrad

All Implemented Interfaces:

java.io.Serializable

public class AdaGrad
extends java.lang.Object
implements java.io.Serializable

Vectorized Learning Rate used per Connection Weight Adapted from: http://xcorr.net/2014/01/23/adagrad-eliminating-learning-rates-in-stochastic-gradient-descent/

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
INDArray	adjustedGradient
double	fudgeFactor
INDArray	gradient
INDArray	historicalGradient
int[]	shape

Constructor Summary

Constructors

Constructor and Description
AdaGrad(int[] shape) Create adagrad with the specified shape
AdaGrad(int rows, int cols) Initializes adagrad with a gamma of 1e-2
AdaGrad(int rows, int cols, double gamma)

Method Summary

Modifier and Type	Method and Description
INDArray	getLearningRates(INDArray gradient) Gets feature specific learning rates Adagrad keeps a history of gradients being passed in.
double	getMasterStepSize()
boolean	isDecayLr()
void	setDecayLr(boolean decayLr)
void	setMasterStepSize(double masterStepSize)

Methods inherited from class java.lang.Object

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

Field Detail

historicalGradient

public INDArray historicalGradient

adjustedGradient

public INDArray adjustedGradient

fudgeFactor

public double fudgeFactor

gradient

public INDArray gradient

shape

public int[] shape

Constructor Detail

AdaGrad

public AdaGrad(int rows,
               int cols,
               double gamma)

AdaGrad

public AdaGrad(int[] shape)

Create adagrad with the specified shape

Parameters:

shape -

AdaGrad

public AdaGrad(int rows,
               int cols)

Initializes adagrad with a gamma of 1e-2

Parameters:

rows - the rows for the gradients

cols - the number of columns for the gradient

Method Detail

getLearningRates

public INDArray getLearningRates(INDArray gradient)

Gets feature specific learning rates Adagrad keeps a history of gradients being passed in. Note that each gradient passed in becomes adapted over time, hence the name adagrad

Parameters:

gradient - the gradient to getFromOrigin learning rates for

Returns:

the feature specific learning rates

getMasterStepSize

public double getMasterStepSize()

setMasterStepSize

public void setMasterStepSize(double masterStepSize)

isDecayLr

public boolean isDecayLr()

setDecayLr

public void setDecayLr(boolean decayLr)