org.nd4j.linalg.learning.config

Class AdaDelta

java.lang.Object

org.nd4j.linalg.learning.config.AdaDelta

All Implemented Interfaces:

Serializable, Cloneable, IUpdater

public class AdaDelta
extends Object
implements IUpdater

http://www.matthewzeiler.com/pubs/googleTR2012/googleTR2012.pdf https://arxiv.org/pdf/1212.5701v1.pdf

Ada delta updater. More robust adagrad that keeps track of a moving window average of the gradient rather than the every decaying learning rates of adagrad

Note: AdaDelta is unique in that it doesn't require manual setting of a learning rate

Author:

Adam Gibson

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	AdaDelta.Builder

Field Summary

Fields

Modifier and Type	Field and Description
static double	DEFAULT_ADADELTA_EPSILON
static double	DEFAULT_ADADELTA_RHO

Constructor Summary

Constructors

Constructor and Description
AdaDelta()

Method Summary

Modifier and Type	Method and Description
void	applySchedules(int iteration, double newLearningRate) Apply the new learning rate and any other schedules
AdaDelta	clone() Clone the updater
GradientUpdater	instantiate(INDArray viewArray, boolean initializeViewArray) Create a new gradient updater
long	stateSize(long numParams) Determine the updater state size for the given number of parameters.

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.nd4j.linalg.learning.config.IUpdater

equals

Field Detail

DEFAULT_ADADELTA_RHO

public static final double DEFAULT_ADADELTA_RHO

See Also:

Constant Field Values

DEFAULT_ADADELTA_EPSILON

public static final double DEFAULT_ADADELTA_EPSILON

See Also:

Constant Field Values

Constructor Detail

AdaDelta

public AdaDelta()

Method Detail

stateSize

public long stateSize(long numParams)

Description copied from interface: IUpdater

Determine the updater state size for the given number of parameters. Usually a integer multiple (0,1 or 2) times the number of parameters in a layer.

Specified by:

stateSize in interface IUpdater

Parameters:

numParams - Number of parameters

Returns:

Updater state size for the given number of parameters

applySchedules

public void applySchedules(int iteration,
                           double newLearningRate)

Description copied from interface: IUpdater

Apply the new learning rate and any other schedules

Specified by:

applySchedules in interface IUpdater

Parameters:

iteration - Current iteration count

newLearningRate - new learning rate to set for the updater

instantiate

public GradientUpdater instantiate(INDArray viewArray,
                                   boolean initializeViewArray)

Description copied from interface: IUpdater

Create a new gradient updater

Specified by:

instantiate in interface IUpdater

Parameters:

viewArray - The updater state size view away

initializeViewArray - If true: initialise the updater state

Returns:

clone

public AdaDelta clone()

Description copied from interface: IUpdater

Clone the updater

Specified by:

clone in interface IUpdater

Overrides:

clone in class Object