org.deeplearning4j.nn.layers.normalization

Class LocalResponseNormalization

java.lang.Object

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

org.deeplearning4j.nn.layers.normalization.LocalResponseNormalization

All Implemented Interfaces:

Serializable, Cloneable, Layer, Model

public class LocalResponseNormalization
extends AbstractLayer<LocalResponseNormalization>

Deep neural net normalization approach normalizes activations between layers "brightness normalization" Used for nets like AlexNet

For a^i_{x,y} the activity of a neuron computed by applying kernel i at position (x,y) and applying ReLU nonlinearity, the response normalized activation b^i_{x,y} is given by:

x^2 = (a^j_{x,y})^2 unitScale = (k + alpha * sum_{j=max(0, i - n/2)}^{max(N-1, i + n/2)} (a^j_{x,y})^2 ) y = b^i_{x,y} = x * unitScale**-beta

gy = epsilon (aka deltas from previous layer) sumPart = sum(a^j_{x,y} * gb^j_{x,y}) gx = gy * unitScale**-beta - 2 * alpha * beta * sumPart/unitScale * a^i_{x,y}

Reference: http://www.cs.toronto.edu/~fritz/absps/imagenet.pdf https://github.com/vlfeat/matconvnet/issues/10 Chainer

Created by nyghtowl on 10/29/15.

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
protected static org.slf4j.Logger	log

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

cacheMode, conf, dropoutApplied, dropoutMask, index, input, iterationListeners, maskArray, maskState, preOutput

Constructor Summary

Constructors

Constructor and Description
LocalResponseNormalization(NeuralNetConfiguration conf)
LocalResponseNormalization(NeuralNetConfiguration conf, org.nd4j.linalg.api.ndarray.INDArray input)

Method Summary

Modifier and Type	Method and Description
org.nd4j.linalg.api.ndarray.INDArray	activate(boolean training) Trigger an activation with the last specified input
org.nd4j.linalg.api.ndarray.INDArray	activationMean() Calculate the mean representation for the activation for this layer
org.nd4j.linalg.primitives.Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon) Calculate the gradient relative to the error in the next layer
Gradient	calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray) Calculate the gradient
double	calcL1(boolean backpropParamsOnly) Calculate the l1 regularization term 0.0 if regularization is not used.
double	calcL2(boolean backpropParamsOnly) Calculate the l2 regularization term 0.0 if regularization is not used.
Layer	clone() Clone the layer
void	fit(org.nd4j.linalg.api.ndarray.INDArray input) Fit the model to the given data
org.nd4j.linalg.api.ndarray.INDArray	getParam(String param) Get the parameter
boolean	isPretrainLayer() Returns true if the layer can be trained in an unsupervised/pretrain manner (VAE, RBMs etc)
void	merge(Layer layer, int batchSize) Averages the given logistic regression from a mini batch into this layer
org.nd4j.linalg.api.ndarray.INDArray	params() Returns the parameters of the neural network as a flattened row vector
org.nd4j.linalg.api.ndarray.INDArray	preOutput(boolean training)
void	setParams(org.nd4j.linalg.api.ndarray.INDArray params) Set the parameters for this model.
Layer	transpose() Return a transposed copy of the weights/bias (this means reverse the number of inputs and outputs on the weights)
Layer.Type	type() Returns the layer type

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

accumulateScore, activate, activate, activate, activate, activate, addListeners, applyDropOutIfNecessary, applyLearningRateScoreDecay, applyMask, batchSize, clear, computeGradientAndScore, conf, derivativeActivation, error, feedForwardMaskArray, fit, getGradientsViewArray, getIndex, getInput, getInputMiniBatchSize, getListeners, getMaskArray, getOptimizer, gradient, gradientAndScore, init, initParams, input, iterate, layerConf, layerId, numParams, numParams, paramTable, paramTable, preOutput, preOutput, preOutput, score, setBackpropGradientsViewArray, setCacheMode, setConf, setIndex, setInput, setInputMiniBatchSize, setListeners, setListeners, setMaskArray, setParam, setParams, setParamsViewArray, setParamTable, update, update, validateInput

Methods inherited from class java.lang.Object

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

Field Detail

log

protected static final org.slf4j.Logger log

Constructor Detail

LocalResponseNormalization

public LocalResponseNormalization(NeuralNetConfiguration conf,
                                  org.nd4j.linalg.api.ndarray.INDArray input)

LocalResponseNormalization

public LocalResponseNormalization(NeuralNetConfiguration conf)

Method Detail

clone

public Layer clone()

Description copied from interface: Layer

Clone the layer

Specified by:

clone in interface Layer

Specified by:

clone in class AbstractLayer<LocalResponseNormalization>

Returns:

calcL2

public double calcL2(boolean backpropParamsOnly)

Description copied from interface: Layer

Calculate the l2 regularization term
0.0 if regularization is not used. Or 0.5 * l2Coeff * l2Magnitude otherwise.
Note that this does not divide by mini-batch size

Specified by:

calcL2 in interface Layer

Overrides:

calcL2 in class AbstractLayer<LocalResponseNormalization>

Parameters:

backpropParamsOnly - If true: calculate L2 based on backprop params only. If false: calculate based on all params (including pretrain params, if any)

Returns:

the l2 regularization term for this layer.

calcL1

public double calcL1(boolean backpropParamsOnly)

Description copied from interface: Layer

Calculate the l1 regularization term
0.0 if regularization is not used. Or l1Coeff * l1Magnitude otherwise.
Note that this does not divide by mini-batch size

Specified by:

calcL1 in interface Layer

Overrides:

calcL1 in class AbstractLayer<LocalResponseNormalization>

Parameters:

backpropParamsOnly - If true: calculate L1 based on backprop params only. If false: calculate based on all params (including pretrain params, if any)

Returns:

the l1 regularization term for this layer.

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<LocalResponseNormalization>

Returns:

fit

public void fit(org.nd4j.linalg.api.ndarray.INDArray input)

Description copied from interface: Model

Fit the model to the given data

Specified by:

fit in interface Model

Overrides:

fit in class AbstractLayer<LocalResponseNormalization>

Parameters:

input - the data to fit the model to

backpropGradient

public org.nd4j.linalg.primitives.Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray> backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)

Description copied from interface: Layer

Calculate the gradient relative to the error in the next layer

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.

Returns:

Pair where Gradient is gradient for this layer, INDArray is epsilon 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() == (w^(L)*(delta^(L))^T)^T

activate

public org.nd4j.linalg.api.ndarray.INDArray activate(boolean training)

Description copied from interface: Layer

Trigger an activation with the last specified input

Parameters:

training - training or test mode

Returns:

the activation of the last specified input

transpose

public Layer transpose()

Description copied from interface: Layer

Return a transposed copy of the weights/bias (this means reverse the number of inputs and outputs on the weights)

Specified by:

transpose in interface Layer

Overrides:

transpose in class AbstractLayer<LocalResponseNormalization>

Returns:

the transposed layer

isPretrainLayer

public boolean isPretrainLayer()

Description copied from interface: Layer

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

Returns:

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

calcGradient

public Gradient calcGradient(Gradient layerError,
                             org.nd4j.linalg.api.ndarray.INDArray indArray)

Description copied from interface: Layer

Calculate the gradient

Specified by:

calcGradient in interface Layer

Overrides:

calcGradient in class AbstractLayer<LocalResponseNormalization>

Parameters:

layerError - the layer error

Returns:

the gradient

merge

public void merge(Layer layer,
                  int batchSize)

Description copied from class: AbstractLayer

Averages the given logistic regression from a mini batch into this layer

Specified by:

merge in interface Layer

Overrides:

merge in class AbstractLayer<LocalResponseNormalization>

Parameters:

layer - the logistic regression layer to average into this layer

batchSize - the batch size

activationMean

public org.nd4j.linalg.api.ndarray.INDArray activationMean()

Description copied from interface: Layer

Calculate the mean representation for the activation for this layer

Returns:

the activation mean for this layer

params

public org.nd4j.linalg.api.ndarray.INDArray params()

Description copied from class: AbstractLayer

Returns the parameters of the neural network as a flattened row vector

Specified by:

params in interface Model

Overrides:

params in class AbstractLayer<LocalResponseNormalization>

Returns:

the parameters of the neural network

getParam

public org.nd4j.linalg.api.ndarray.INDArray getParam(String param)

Description copied from interface: Model

Get the parameter

Specified by:

getParam in interface Model

Overrides:

getParam in class AbstractLayer<LocalResponseNormalization>

Parameters:

param - the key of the parameter

Returns:

the parameter vector/matrix with that particular key

setParams

public void setParams(org.nd4j.linalg.api.ndarray.INDArray params)

Description copied from interface: Model

Set the parameters for this model. This expects a linear ndarray which then be unpacked internally relative to the expected ordering of the model

Specified by:

setParams in interface Model

Overrides:

setParams in class AbstractLayer<LocalResponseNormalization>

Parameters:

params - the parameters for the model

preOutput

public org.nd4j.linalg.api.ndarray.INDArray preOutput(boolean training)

Specified by:

preOutput in class AbstractLayer<LocalResponseNormalization>