org.deeplearning4j.nn.conf.constraint

Class BaseConstraint

java.lang.Object

org.deeplearning4j.nn.conf.constraint.BaseConstraint

All Implemented Interfaces:

Serializable, Cloneable, LayerConstraint

Direct Known Subclasses:

MaxNormConstraint, MinMaxNormConstraint, NonNegativeConstraint, UnitNormConstraint

public abstract class BaseConstraint
extends Object
implements LayerConstraint

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static double	DEFAULT_EPSILON
protected int[]	dimensions
protected double	epsilon
protected Set<String>	params

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	BaseConstraint()
protected	BaseConstraint(Set<String> paramNames, int... dimensions)

Method Summary

Modifier and Type	Method and Description
abstract void	apply(org.nd4j.linalg.api.ndarray.INDArray param)
void	applyConstraint(Layer layer, int iteration, int epoch) Apply a given constraint to a layer at each iteration in the provided epoch, after parameters have been updated.
abstract BaseConstraint	clone()
static int[]	getBroadcastDims(int[] reduceDimensions, int rank)

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.deeplearning4j.nn.api.layers.LayerConstraint

getParams, setParams

Field Detail

DEFAULT_EPSILON

public static final double DEFAULT_EPSILON

See Also:

Constant Field Values

params

protected Set<String> params

epsilon

protected double epsilon

dimensions

protected int[] dimensions

Constructor Detail

BaseConstraint

protected BaseConstraint()

BaseConstraint

protected BaseConstraint(Set<String> paramNames,
                         int... dimensions)

Method Detail

applyConstraint

public void applyConstraint(Layer layer,
                            int iteration,
                            int epoch)

Description copied from interface: LayerConstraint

Apply a given constraint to a layer at each iteration in the provided epoch, after parameters have been updated.

Specified by:

applyConstraint in interface LayerConstraint

Parameters:

layer - org.deeplearning4j.nn.api.Layer

iteration - given iteration as integer

epoch - current epoch as integer

apply

public abstract void apply(org.nd4j.linalg.api.ndarray.INDArray param)

clone

public abstract BaseConstraint clone()

Specified by:

clone in interface LayerConstraint

Overrides:

clone in class Object

getBroadcastDims

public static int[] getBroadcastDims(int[] reduceDimensions,
                                     int rank)