org.deeplearning4j.nn.layers.normalization

Class BatchNormalization

java.lang.Object

org.deeplearning4j.nn.layers.AbstractLayer<LayerConfT>

org.deeplearning4j.nn.layers.BaseLayer<BatchNormalization>

org.deeplearning4j.nn.layers.normalization.BatchNormalization

All Implemented Interfaces:

Serializable, Cloneable, Layer, Model, Trainable

public class BatchNormalization
extends BaseLayer<BatchNormalization>

See Also:

Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from interface org.deeplearning4j.nn.api.Layer

Layer.TrainingMode, Layer.Type

Field Summary

Fields

Modifier and Type	Field and Description
static String	BATCH_NORM_CUDNN_HELPER_CLASS_NAME
protected int	helperCountFail
protected int	index
protected List<TrainingListener>	listeners
protected static double	ONE_ON_2LOGE_10
protected INDArray	std
protected INDArray	xHat
protected INDArray	xMu

Fields inherited from class org.deeplearning4j.nn.layers.BaseLayer

gradient, gradientsFlattened, gradientViews, optimizer, params, paramsFlattened, score, solver, weightNoiseParams

Fields inherited from class org.deeplearning4j.nn.layers.AbstractLayer

cacheMode, conf, dataType, dropoutApplied, epochCount, input, inputModificationAllowed, iterationCount, maskArray, maskState, preOutput, trainingListeners

Constructor Summary

Constructors

Constructor and Description
BatchNormalization(NeuralNetConfiguration conf, DataType dataType)

Method Summary

Modifier and Type	Method and Description
INDArray	activate(boolean training, LayerWorkspaceMgr workspaceMgr) Perform forward pass and return the activations array with the last set input
Pair<Gradient,INDArray>	backpropGradient(INDArray epsilon, LayerWorkspaceMgr workspaceMgr) Calculate the gradient relative to the error in the next layer
void	fit(INDArray input, LayerWorkspaceMgr workspaceMgr) Fit the model to the given data
LayerHelper	getHelper()
int	getIndex() Get the layer index.
Collection<TrainingListener>	getListeners() Get the iteration listeners for this layer.
long[]	getShape(INDArray x)
Gradient	gradient() Get the gradient.
boolean	isPretrainLayer() Returns true if the layer can be trained in an unsupervised/pretrain manner (AE, VAE, etc)
INDArray	preOutput(INDArray x, Layer.TrainingMode training, LayerWorkspaceMgr workspaceMgr)
void	setIndex(int index) Set the layer index.
void	setListeners(TrainingListener... listeners) Set the TrainingListeners for this model.
Layer.Type	type() Returns the layer type
boolean	updaterDivideByMinibatch(String paramName) DL4J layers typically produce the sum of the gradients during the backward pass for each layer, and if required (if minibatch=true) then divide by the minibatch size. However, there are some exceptions, such as the batch norm mean/variance estimate parameters: these "gradients" are actually not gradients, but are updates to be applied directly to the parameter vector.

Methods inherited from class org.deeplearning4j.nn.layers.BaseLayer

calcRegularizationScore, clear, clearNoiseWeightParams, clone, computeGradientAndScore, fit, getGradientsViewArray, getOptimizer, getParam, getParamWithNoise, hasBias, hasLayerNorm, layerConf, numParams, params, paramTable, paramTable, preOutput, preOutputWithPreNorm, score, setBackpropGradientsViewArray, setParam, setParams, setParams, setParamsViewArray, setParamTable, setScoreWithZ, toString, update, update

Methods inherited from class org.deeplearning4j.nn.layers.AbstractLayer

activate, addListeners, allowInputModification, applyConstraints, applyDropOutIfNecessary, applyMask, assertInputSet, backpropDropOutIfPresent, batchSize, close, conf, feedForwardMaskArray, getConfig, getEpochCount, getInput, getInputMiniBatchSize, getMaskArray, gradientAndScore, init, input, layerId, numParams, setCacheMode, setConf, setEpochCount, setInput, setInputMiniBatchSize, setListeners, setMaskArray

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.deeplearning4j.nn.api.Layer

getIterationCount, setIterationCount

Field Detail

ONE_ON_2LOGE_10

protected static final double ONE_ON_2LOGE_10

helperCountFail

protected int helperCountFail

index

protected int index

listeners

protected List<TrainingListener> listeners

std

protected INDArray std

xMu

protected INDArray xMu

xHat

protected INDArray xHat

BATCH_NORM_CUDNN_HELPER_CLASS_NAME

public static final String BATCH_NORM_CUDNN_HELPER_CLASS_NAME

See Also:

Constant Field Values

Constructor Detail

BatchNormalization

public BatchNormalization(NeuralNetConfiguration conf,
                          DataType dataType)

Method Detail

type

public Layer.Type type()

Description copied from interface: Layer

Returns the layer type

Specified by:

type in interface Layer

Overrides:

type in class AbstractLayer<BatchNormalization>

Returns:

backpropGradient

public Pair<Gradient,INDArray> backpropGradient(INDArray epsilon,
                                                LayerWorkspaceMgr workspaceMgr)

Description copied from interface: Layer

Calculate the gradient relative to the error in the next layer

Specified by:

backpropGradient in interface Layer

Overrides:

backpropGradient in class BaseLayer<BatchNormalization>

Parameters:

epsilon - w^(L+1)*delta^(L+1). Or, equiv: dC/da, i.e., (dC/dz)*(dz/da) = dC/da, where C is cost function a=sigma(z) is activation.

workspaceMgr - Workspace manager

Returns:

Pair where Gradient is gradient for this layer, INDArray is epsilon (activation gradient) needed by next layer, but before element-wise multiply by sigmaPrime(z). So for standard feed-forward layer, if this layer is L, then return.getSecond() == dL/dIn = (w^(L)*(delta^(L))^T)^T. Note that the returned array should be placed in the ArrayType.ACTIVATION_GRAD workspace via the workspace manager

fit

public void fit(INDArray input,
                LayerWorkspaceMgr workspaceMgr)

Description copied from interface: Model

Fit the model to the given data

Specified by:

fit in interface Model

Overrides:

fit in class BaseLayer<BatchNormalization>

Parameters:

input - the data to fit the model to

activate

public INDArray activate(boolean training,
                         LayerWorkspaceMgr workspaceMgr)

Description copied from interface: Layer

Perform forward pass and return the activations array with the last set input

Specified by:

activate in interface Layer

Overrides:

activate in class BaseLayer<BatchNormalization>

Parameters:

training - training or test mode

workspaceMgr - Workspace manager

Returns:

the activation (layer output) of the last specified input. Note that the returned array should be placed in the ArrayType.ACTIVATIONS workspace via the workspace manager

gradient

public Gradient gradient()

Description copied from interface: Model

Get the gradient. Note that this method will not calculate the gradient, it will rather return the gradient that has been computed before. For calculating the gradient, see Model.computeGradientAndScore(LayerWorkspaceMgr) } .

Specified by:

gradient in interface Model

Overrides:

gradient in class BaseLayer<BatchNormalization>

Returns:

the gradient for this model, as calculated before

preOutput

public INDArray preOutput(INDArray x,
                          Layer.TrainingMode training,
                          LayerWorkspaceMgr workspaceMgr)

getListeners

public Collection<TrainingListener> getListeners()

Description copied from interface: Layer

Get the iteration listeners for this layer.

Specified by:

getListeners in interface Layer

Overrides:

getListeners in class AbstractLayer<BatchNormalization>

setListeners

public void setListeners(TrainingListener... listeners)

Description copied from interface: Layer

Set the TrainingListeners for this model. If any listeners have previously been set, they will be replaced by this method

Specified by:

setListeners in interface Layer

Specified by:

setListeners in interface Model

Overrides:

setListeners in class AbstractLayer<BatchNormalization>

setIndex

public void setIndex(int index)

Description copied from interface: Layer

Set the layer index.

Specified by:

setIndex in interface Layer

Overrides:

setIndex in class AbstractLayer<BatchNormalization>

getIndex

public int getIndex()

Description copied from interface: Layer

Get the layer index.

Specified by:

getIndex in interface Layer

Overrides:

getIndex in class AbstractLayer<BatchNormalization>

isPretrainLayer

public boolean isPretrainLayer()

Description copied from interface: Layer

Returns true if the layer can be trained in an unsupervised/pretrain manner (AE, VAE, etc)

Returns:

true if the layer can be pretrained (using fit(INDArray), false otherwise

getHelper

public LayerHelper getHelper()

Specified by:

getHelper in interface Layer

Overrides:

getHelper in class AbstractLayer<BatchNormalization>

Returns:

Get the layer helper, if any

getShape

public long[] getShape(INDArray x)

updaterDivideByMinibatch

public boolean updaterDivideByMinibatch(String paramName)

Description copied from interface: Trainable

DL4J layers typically produce the sum of the gradients during the backward pass for each layer, and if required (if minibatch=true) then divide by the minibatch size.
However, there are some exceptions, such as the batch norm mean/variance estimate parameters: these "gradients" are actually not gradients, but are updates to be applied directly to the parameter vector. Put another way, most gradients should be divided by the minibatch to get the average; some "gradients" are actually final updates already, and should not be divided by the minibatch size.

Specified by:

updaterDivideByMinibatch in interface Trainable

Overrides:

updaterDivideByMinibatch in class AbstractLayer<BatchNormalization>

Parameters:

paramName - Name of the parameter

Returns:

True if gradients should be divided by minibatch (most params); false otherwise (edge cases like batch norm mean/variance estimates)