Class Nadam

java.lang.Object

org.tensorflow.framework.optimizers.Optimizer

org.tensorflow.framework.optimizers.Nadam

public class Nadam
extends Optimizer

Nadam Optimizer that implements the NAdam algorithm.

Much like Adam is essentially RMSprop with momentum, Nadam is Adam with Nesterov momentum.

See Also:

• Dozat, 2015

Nested Class Summary

Nested classes/interfaces inherited from class Optimizer

Optimizer.GradAndVar<T>, Optimizer.Options

Field Summary

Fields

Modifier and Type	Field	Description
static final float	BETA_ONE_DEFAULT
static final float	BETA_TWO_DEFAULT
static final float	EPSILON_DEFAULT
static final String	FIRST_MOMENT
static final float	LEARNING_RATE_DEFAULT
static final String	MOMENTUM
static final String	SECOND_MOMENT

Fields inherited from class Optimizer

globals, graph, tf, VARIABLE_V2

Constructor Summary

Constructors

Constructor	Description
Nadam(Graph graph)	Creates a Nadam Optimizer
Nadam(Graph graph, float learningRate)	Creates a Nadam Optimizer
Nadam(Graph graph, float learningRate, float betaOne, float betaTwo, float epsilon)	Creates a Nadam Optimizer
Nadam(Graph graph, String name, float learningRate)	Creates a Nadam Optimizer
Nadam(Graph graph, String name, float learningRate, float betaOne, float betaTwo, float epsilon)	Creates a Nadam Optimizer

Method Summary

Modifier and Type	Method	Description
protected <T extends TType> Op	applyDense(Ops deps, Output<T> gradient, Output<T> variable)	Generates the gradient update operations for the specific variable and gradient.
protected void	createSlots(List<Output<? extends TType>> variables)	Performs a No-op slot creation method.
protected Op	finish(List<Op> updateOperations, String name)	Gathers up the update operations into a single op that can be used as a run target.
String	getOptimizerName()	Get the Name of the optimizer.
protected Optional<Op>	prepare(String scopeName)	Returns a No-op prepare.

Methods inherited from class Optimizer

applyGradients, computeGradients, createName, createSlot, getSlot, getTF, minimize, minimize

Methods inherited from class Object

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

Field Details

LEARNING_RATE_DEFAULT

public static final float LEARNING_RATE_DEFAULT

See Also:

• Constant Field Values

EPSILON_DEFAULT

public static final float EPSILON_DEFAULT

See Also:

• Constant Field Values

BETA_ONE_DEFAULT

public static final float BETA_ONE_DEFAULT

See Also:

• Constant Field Values

BETA_TWO_DEFAULT

public static final float BETA_TWO_DEFAULT

See Also:

• Constant Field Values

FIRST_MOMENT

public static final String FIRST_MOMENT

See Also:

• Constant Field Values

SECOND_MOMENT

public static final String SECOND_MOMENT

See Also:

• Constant Field Values

MOMENTUM

public static final String MOMENTUM

See Also:

• Constant Field Values

Constructor Details

Nadam

public Nadam(Graph graph)

Creates a Nadam Optimizer

Parameters:

graph - the TensorFlow graph

Nadam

public Nadam(Graph graph,
 float learningRate)

Creates a Nadam Optimizer

Parameters:

graph - the TensorFlow graph

learningRate - the learning rate, defaults to 0.001

Nadam

public Nadam(Graph graph,
 float learningRate,
 float betaOne,
 float betaTwo,
 float epsilon)

Creates a Nadam Optimizer

Parameters:

graph - the TensorFlow graph

learningRate - the learning rate, defaults to 0.001

betaOne - The exponential decay rate for the 1st moment estimates. Default is 0.9.

betaTwo - The exponential decay rate for the exponentially weighted infinity norm. Default is 0.999.

epsilon - A small constant for numerical stability. Default is 1e-8.

Nadam

public Nadam(Graph graph,
 String name,
 float learningRate)

Creates a Nadam Optimizer

Parameters:

graph - the TensorFlow graph

name - the name for this Optimizer, defaults to "Nadam"

learningRate - the learning rate, defaults to 0.001

Nadam

public Nadam(Graph graph,
 String name,
 float learningRate,
 float betaOne,
 float betaTwo,
 float epsilon)

Creates a Nadam Optimizer

Parameters:

graph - the TensorFlow graph

name - the name for this Optimizer, defaults to "Nadam"

learningRate - the learning rate, defaults to 0.001

betaOne - The exponential decay rate for the 1st moment estimates. Default is 0.9.

betaTwo - The exponential decay rate for the exponentially weighted infinity norm. Default is 0.999.

epsilon - A small constant for numerical stability. Default is 1e-8.

Method Details

createSlots

protected void createSlots(List<Output<? extends TType>> variables)

Performs a No-op slot creation method.

Overrides:

createSlots in class Optimizer

Parameters:

variables - The variables to create slots for.

prepare

protected Optional<Op> prepare(String scopeName)

Returns a No-op prepare.

Overrides:

prepare in class Optimizer

Parameters:

scopeName - The scope name to use for any variable creations.

Returns:

a No-op to prepare this optimizer, or empty if none.

applyDense

protected <T extends TType> Op applyDense(Ops deps,
 Output<T> gradient,
 Output<T> variable)

Generates the gradient update operations for the specific variable and gradient.

Specified by:

applyDense in class Optimizer

Type Parameters:

T - The type of the variable.

Parameters:

gradient - The gradient to use.

variable - The variable to update.

Returns:

An operand which applies the desired optimizer update to the variable.

finish

protected Op finish(List<Op> updateOperations,
 String name)

Gathers up the update operations into a single op that can be used as a run target.

Adds the betaOne, betaTwo and mu updates to the end of the updates list.

Overrides:

finish in class Optimizer

Parameters:

updateOperations - The update operations.

name - The name of the run target.

Returns:

A NoOp with a control dependency on each update operation.

getOptimizerName

public String getOptimizerName()

Get the Name of the optimizer.

Specified by:

getOptimizerName in class Optimizer

Returns:

The optimizer name.