Uses of Interface
org.deeplearning4j.nn.gradient.Gradient

Packages that use Gradient

Package	Description
org.deeplearning4j.nn.api
org.deeplearning4j.nn.api.layers
org.deeplearning4j.nn.gradient
org.deeplearning4j.nn.graph
org.deeplearning4j.nn.graph.vertex
org.deeplearning4j.nn.graph.vertex.impl
org.deeplearning4j.nn.graph.vertex.impl.rnn
org.deeplearning4j.nn.layers
org.deeplearning4j.nn.layers.convolution
org.deeplearning4j.nn.layers.convolution.subsampling
org.deeplearning4j.nn.layers.feedforward.embedding
org.deeplearning4j.nn.layers.feedforward.rbm
org.deeplearning4j.nn.layers.normalization
org.deeplearning4j.nn.layers.recurrent
org.deeplearning4j.nn.multilayer
org.deeplearning4j.nn.updater
org.deeplearning4j.nn.updater.graph
org.deeplearning4j.optimize.api
org.deeplearning4j.optimize.solvers

Uses of Gradient in org.deeplearning4j.nn.api

Methods in org.deeplearning4j.nn.api that return Gradient

Modifier and Type	Method and Description
Gradient	Layer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray) Calculate the gradient
Gradient	Layer.error(org.nd4j.linalg.api.ndarray.INDArray input) Calculate error with respect to the current layer.
Gradient	Model.gradient() Calculate a gradient

Methods in org.deeplearning4j.nn.api that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	Layer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon) Calculate the gradient relative to the error in the next layer
Pair<Gradient,Double>	Model.gradientAndScore() Get the gradient and score

Methods in org.deeplearning4j.nn.api with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	Layer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray) Calculate the gradient
void	Model.update(Gradient gradient) Update layer weights and biases with gradient change
void	Updater.update(Layer layer, Gradient gradient, int iteration, int miniBatchSize) Updater: updates the model

Uses of Gradient in org.deeplearning4j.nn.api.layers

Methods in org.deeplearning4j.nn.api.layers that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	RecurrentLayer.tbpttBackpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon, int tbpttBackLength) Truncated BPTT equivalent of Layer.backpropGradient().

Uses of Gradient in org.deeplearning4j.nn.gradient

Classes in org.deeplearning4j.nn.gradient that implement Gradient

Modifier and Type	Class and Description
class	DefaultGradient Default gradient implementation.

Uses of Gradient in org.deeplearning4j.nn.graph

Fields in org.deeplearning4j.nn.graph declared as Gradient

Modifier and Type	Field and Description
protected Gradient	ComputationGraph.gradient

Methods in org.deeplearning4j.nn.graph that return Gradient

Modifier and Type	Method and Description
Gradient	ComputationGraph.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray... epsilons) Calculate the gradient of the network with respect to some external errors.
Gradient	ComputationGraph.gradient()

Methods in org.deeplearning4j.nn.graph that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,Double>	ComputationGraph.gradientAndScore()

Methods in org.deeplearning4j.nn.graph with parameters of type Gradient

Modifier and Type	Method and Description
void	ComputationGraph.update(Gradient gradient)

Uses of Gradient in org.deeplearning4j.nn.graph.vertex

Methods in org.deeplearning4j.nn.graph.vertex that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	GraphVertex.doBackward(boolean tbptt) Do backward pass

Uses of Gradient in org.deeplearning4j.nn.graph.vertex.impl

Methods in org.deeplearning4j.nn.graph.vertex.impl that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	StackVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	LayerVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	SubsetVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	PreprocessorVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	UnstackVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	MergeVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	ElementWiseVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	InputVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	L2Vertex.doBackward(boolean tbptt)

Uses of Gradient in org.deeplearning4j.nn.graph.vertex.impl.rnn

Methods in org.deeplearning4j.nn.graph.vertex.impl.rnn that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	LastTimeStepVertex.doBackward(boolean tbptt)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray[]>	DuplicateToTimeSeriesVertex.doBackward(boolean tbptt)

Uses of Gradient in org.deeplearning4j.nn.layers

Fields in org.deeplearning4j.nn.layers declared as Gradient

Modifier and Type	Field and Description
protected Gradient	BaseLayer.gradient

Methods in org.deeplearning4j.nn.layers that return Gradient

Modifier and Type	Method and Description
Gradient	LossLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
Gradient	BaseLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
Gradient	ActivationLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
protected Gradient	BaseLayer.createGradient(org.nd4j.linalg.api.ndarray.INDArray... gradients) Create a gradient list based on the passed in parameters.
protected Gradient	BasePretrainNetwork.createGradient(org.nd4j.linalg.api.ndarray.INDArray wGradient, org.nd4j.linalg.api.ndarray.INDArray vBiasGradient, org.nd4j.linalg.api.ndarray.INDArray hBiasGradient)
Gradient	BaseLayer.error(org.nd4j.linalg.api.ndarray.INDArray errorSignal)
Gradient	LossLayer.gradient() Gets the gradient from one training iteration
Gradient	BaseLayer.gradient()
Gradient	BaseOutputLayer.gradient() Gets the gradient from one training iteration

Methods in org.deeplearning4j.nn.layers that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	BasePretrainNetwork.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	LossLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	BaseLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	BaseOutputLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	ActivationLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,Double>	LossLayer.gradientAndScore()
Pair<Gradient,Double>	BaseLayer.gradientAndScore()
Pair<Gradient,Double>	BaseOutputLayer.gradientAndScore()

Methods in org.deeplearning4j.nn.layers with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	LossLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
Gradient	BaseLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
Gradient	ActivationLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
void	BaseLayer.update(Gradient gradient)

Uses of Gradient in org.deeplearning4j.nn.layers.convolution

Methods in org.deeplearning4j.nn.layers.convolution that return Gradient

Modifier and Type	Method and Description
Gradient	ConvolutionLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)

Methods in org.deeplearning4j.nn.layers.convolution that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	ConvolutionLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	ConvolutionHelper.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray input, org.nd4j.linalg.api.ndarray.INDArray weights, org.nd4j.linalg.api.ndarray.INDArray delta, int[] kernel, int[] strides, int[] pad, org.nd4j.linalg.api.ndarray.INDArray biasGradView, org.nd4j.linalg.api.ndarray.INDArray weightGradView, String afn, ConvolutionLayer.AlgoMode mode, ConvolutionMode convolutionMode)

Methods in org.deeplearning4j.nn.layers.convolution with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	ConvolutionLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)

Uses of Gradient in org.deeplearning4j.nn.layers.convolution.subsampling

Methods in org.deeplearning4j.nn.layers.convolution.subsampling that return Gradient

Modifier and Type	Method and Description
Gradient	SubsamplingLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
Gradient	SubsamplingLayer.error(org.nd4j.linalg.api.ndarray.INDArray input)

Methods in org.deeplearning4j.nn.layers.convolution.subsampling that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	SubsamplingLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	SubsamplingHelper.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray input, org.nd4j.linalg.api.ndarray.INDArray epsilon, int[] kernel, int[] strides, int[] pad, SubsamplingLayer.PoolingType poolingType, ConvolutionMode convolutionMode)

Methods in org.deeplearning4j.nn.layers.convolution.subsampling with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	SubsamplingLayer.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)

Uses of Gradient in org.deeplearning4j.nn.layers.feedforward.embedding

Methods in org.deeplearning4j.nn.layers.feedforward.embedding that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	EmbeddingLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)

Uses of Gradient in org.deeplearning4j.nn.layers.feedforward.rbm

Methods in org.deeplearning4j.nn.layers.feedforward.rbm that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	RBM.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)

Uses of Gradient in org.deeplearning4j.nn.layers.normalization

Methods in org.deeplearning4j.nn.layers.normalization that return Gradient

Modifier and Type	Method and Description
Gradient	BatchNormalization.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
Gradient	LocalResponseNormalization.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
Gradient	BatchNormalization.error(org.nd4j.linalg.api.ndarray.INDArray input)
Gradient	BatchNormalization.gradient()

Methods in org.deeplearning4j.nn.layers.normalization that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	BatchNormalization.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	LocalResponseNormalization.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	LocalResponseNormalizationHelper.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray input, org.nd4j.linalg.api.ndarray.INDArray epsilon, double k, double n, double alpha, double beta)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	BatchNormalizationHelper.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray input, org.nd4j.linalg.api.ndarray.INDArray epsilon, int[] shape, org.nd4j.linalg.api.ndarray.INDArray gamma, org.nd4j.linalg.api.ndarray.INDArray dGammaView, org.nd4j.linalg.api.ndarray.INDArray dBetaView, double eps)

Methods in org.deeplearning4j.nn.layers.normalization with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	BatchNormalization.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)
Gradient	LocalResponseNormalization.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray indArray)

Uses of Gradient in org.deeplearning4j.nn.layers.recurrent

Methods in org.deeplearning4j.nn.layers.recurrent that return Gradient

Modifier and Type	Method and Description
Gradient	GravesLSTM.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
Gradient	GravesBidirectionalLSTM.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
Gradient	GravesLSTM.gradient()
Gradient	GravesBidirectionalLSTM.gradient()

Methods in org.deeplearning4j.nn.layers.recurrent that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	GravesLSTM.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	GravesBidirectionalLSTM.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	RnnOutputLayer.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
static Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	LSTMHelpers.backpropGradientHelper(NeuralNetConfiguration conf, org.nd4j.linalg.api.ndarray.INDArray input, org.nd4j.linalg.api.ndarray.INDArray recurrentWeights, org.nd4j.linalg.api.ndarray.INDArray inputWeights, org.nd4j.linalg.api.ndarray.INDArray epsilon, boolean truncatedBPTT, int tbpttBackwardLength, FwdPassReturn fwdPass, boolean forwards, String inputWeightKey, String recurrentWeightKey, String biasWeightKey, Map<String,org.nd4j.linalg.api.ndarray.INDArray> gradientViews)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	GravesLSTM.tbpttBackpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon, int tbpttBackwardLength)
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	GravesBidirectionalLSTM.tbpttBackpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon, int tbpttBackwardLength)

Methods in org.deeplearning4j.nn.layers.recurrent with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	GravesLSTM.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
Gradient	GravesBidirectionalLSTM.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)

Uses of Gradient in org.deeplearning4j.nn.multilayer

Fields in org.deeplearning4j.nn.multilayer declared as Gradient

Modifier and Type	Field and Description
protected Gradient	MultiLayerNetwork.gradient

Methods in org.deeplearning4j.nn.multilayer that return Gradient

Modifier and Type	Method and Description
Gradient	MultiLayerNetwork.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
Gradient	MultiLayerNetwork.error(org.nd4j.linalg.api.ndarray.INDArray errorSignal)
Gradient	MultiLayerNetwork.gradient()

Methods in org.deeplearning4j.nn.multilayer that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	MultiLayerNetwork.backpropGradient(org.nd4j.linalg.api.ndarray.INDArray epsilon)
protected Pair<Gradient,org.nd4j.linalg.api.ndarray.INDArray>	MultiLayerNetwork.calcBackpropGradients(org.nd4j.linalg.api.ndarray.INDArray epsilon, boolean withOutputLayer) Calculate gradients and errors.
Pair<Gradient,Double>	MultiLayerNetwork.gradientAndScore()

Methods in org.deeplearning4j.nn.multilayer with parameters of type Gradient

Modifier and Type	Method and Description
Gradient	MultiLayerNetwork.calcGradient(Gradient layerError, org.nd4j.linalg.api.ndarray.INDArray activation)
void	MultiLayerNetwork.update(Gradient gradient)

Uses of Gradient in org.deeplearning4j.nn.updater

Methods in org.deeplearning4j.nn.updater with parameters of type Gradient

Modifier and Type	Method and Description
void	LayerUpdater.preApply(Layer layer, Gradient gradient, int iteration) Apply gradient normalization: scale based on L2, clipping etc.
void	LayerUpdater.update(Layer layer, Gradient gradient, int iteration, int miniBatchSize)
void	MultiLayerUpdater.update(Layer layer, Gradient gradient, int iteration, int batchSize)

Uses of Gradient in org.deeplearning4j.nn.updater.graph

Methods in org.deeplearning4j.nn.updater.graph with parameters of type Gradient

Modifier and Type	Method and Description
void	ComputationGraphUpdater.update(ComputationGraph graph, Gradient gradient, int iteration, int batchSize) Update the gradients for the given ComputationGraph

Uses of Gradient in org.deeplearning4j.optimize.api

Methods in org.deeplearning4j.optimize.api that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,Double>	ConvexOptimizer.gradientAndScore() The gradient and score for this optimizer

Methods in org.deeplearning4j.optimize.api with parameters of type Gradient

Modifier and Type	Method and Description
void	ConvexOptimizer.updateGradientAccordingToParams(Gradient gradient, Model model, int batchSize) Update the gradient according to the configuration such as adagrad, momentum, and sparsity

Method parameters in org.deeplearning4j.optimize.api with type arguments of type Gradient

Modifier and Type	Method and Description
void	ConvexOptimizer.setupSearchState(Pair<Gradient,Double> pair) Based on the gradient and score setup a search state

Uses of Gradient in org.deeplearning4j.optimize.solvers

Methods in org.deeplearning4j.optimize.solvers that return types with arguments of type Gradient

Modifier and Type	Method and Description
Pair<Gradient,Double>	BaseOptimizer.gradientAndScore()

Methods in org.deeplearning4j.optimize.solvers with parameters of type Gradient

Modifier and Type	Method and Description
void	BaseOptimizer.updateGradientAccordingToParams(Gradient gradient, Model model, int batchSize)

Method parameters in org.deeplearning4j.optimize.solvers with type arguments of type Gradient

Modifier and Type	Method and Description
void	BaseOptimizer.setupSearchState(Pair<Gradient,Double> pair) Setup the initial search state
void	LBFGS.setupSearchState(Pair<Gradient,Double> pair)