Uses of Interface
org.deeplearning4j.nn.weights.IWeightInit

Packages that use IWeightInit

Package	Description
org.deeplearning4j.nn.conf
org.deeplearning4j.nn.conf.layers
org.deeplearning4j.nn.conf.layers.samediff
org.deeplearning4j.nn.params
org.deeplearning4j.nn.transferlearning
org.deeplearning4j.nn.weights
org.deeplearning4j.nn.weights.embeddings

Uses of IWeightInit in org.deeplearning4j.nn.conf

Fields in org.deeplearning4j.nn.conf declared as IWeightInit

Modifier and Type	Field and Description
protected IWeightInit	NeuralNetConfiguration.Builder.weightInitFn

Methods in org.deeplearning4j.nn.conf with parameters of type IWeightInit

Modifier and Type	Method and Description
NeuralNetConfiguration.Builder	NeuralNetConfiguration.Builder.weightInit(IWeightInit weightInit) Weight initialization scheme to use, for initial weight values Note: values set by this method will be applied to all applicable layers in the network, unless a different value is explicitly set on a given layer.

Uses of IWeightInit in org.deeplearning4j.nn.conf.layers

Fields in org.deeplearning4j.nn.conf.layers declared as IWeightInit

Modifier and Type	Field and Description
protected IWeightInit	BaseLayer.weightInitFn
protected IWeightInit	BaseLayer.Builder.weightInitFn Weight initialization scheme to use, for initial weight values
protected IWeightInit	BaseRecurrentLayer.weightInitFnRecurrent
protected IWeightInit	BaseRecurrentLayer.Builder.weightInitFnRecurrent Set the weight initialization for the recurrent weights.

Methods in org.deeplearning4j.nn.conf.layers with parameters of type IWeightInit

Modifier and Type	Method and Description
void	EmbeddingSequenceLayer.Builder.setWeightInitFn(IWeightInit weightInit)
EmbeddingSequenceLayer.Builder	EmbeddingSequenceLayer.Builder.weightInit(IWeightInit weightInit)
EmbeddingLayer.Builder	EmbeddingLayer.Builder.weightInit(IWeightInit weightInit)
T	BaseLayer.Builder.weightInit(IWeightInit weightInit) Weight initialization scheme to use, for initial weight values
T	BaseRecurrentLayer.Builder.weightInitRecurrent(IWeightInit weightInit) Set the weight initialization for the recurrent weights.

Uses of IWeightInit in org.deeplearning4j.nn.conf.layers.samediff

Fields in org.deeplearning4j.nn.conf.layers.samediff with type parameters of type IWeightInit

Modifier and Type	Field and Description
protected Map<String,IWeightInit>	SameDiffLayer.paramWeightInit
protected Map<String,IWeightInit>	SameDiffLayer.Builder.paramWeightInit

Methods in org.deeplearning4j.nn.conf.layers.samediff with parameters of type IWeightInit

Modifier and Type	Method and Description
T	SameDiffLayer.Builder.weightInit(String param, IWeightInit weightInit)

Uses of IWeightInit in org.deeplearning4j.nn.params

Methods in org.deeplearning4j.nn.params with parameters of type IWeightInit

Modifier and Type	Method and Description
protected INDArray	DefaultParamInitializer.createWeightMatrix(long nIn, long nOut, IWeightInit weightInit, INDArray weightParamView, boolean initializeParameters)
protected INDArray	ElementWiseParamInitializer.createWeightMatrix(long nIn, long nOut, IWeightInit weightInit, INDArray weightParamView, boolean initializeParameters)

Uses of IWeightInit in org.deeplearning4j.nn.transferlearning

Fields in org.deeplearning4j.nn.transferlearning declared as IWeightInit

Modifier and Type	Field and Description
protected IWeightInit	FineTuneConfiguration.weightInitFn

Methods in org.deeplearning4j.nn.transferlearning with parameters of type IWeightInit

Modifier and Type	Method and Description
TransferLearning.Builder	TransferLearning.Builder.nInReplace(int layerNum, int nIn, IWeightInit scheme) Modify the architecture of a vertex layer by changing nIn of the specified layer. Note that only the specified layer will be modified - all other layers will not be changed by this call.
TransferLearning.GraphBuilder	TransferLearning.GraphBuilder.nInReplace(String layerName, int nIn, IWeightInit scheme) Modify the architecture of a vertex layer by changing nIn of the specified layer. Note that only the specified layer will be modified - all other layers will not be changed by this call.
TransferLearning.Builder	TransferLearning.Builder.nOutReplace(int layerNum, int nOut, IWeightInit scheme, IWeightInit schemeNext) Modify the architecture of a layer by changing nOut Note this will also affect the layer that follows the layer specified, unless it is the output layer Can specify different weight init schemes for the specified layer and the layer that follows it.
FineTuneConfiguration.Builder	FineTuneConfiguration.Builder.weightInit(IWeightInit weightInit) Weight initialization scheme to use, for initial weight values

Uses of IWeightInit in org.deeplearning4j.nn.weights

Classes in org.deeplearning4j.nn.weights that implement IWeightInit

Modifier and Type	Class and Description
class	WeightInitConstant Initialize to a constant value (deafult 0).
class	WeightInitDistribution Sample weights from a provided Distribution Note that Distribution is not extendable as it is interpreted through Distributions.createDistribution(Distribution).
class	WeightInitIdentity Weights are set to an identity matrix.
class	WeightInitLecunUniform Uniform U[-a,a] with a=3/sqrt(fanIn).
class	WeightInitNormal Normal/Gaussian distribution, with mean 0 and standard deviation 1/sqrt(fanIn).
class	WeightInitRelu : He et al.
class	WeightInitReluUniform He et al.
class	WeightInitSigmoidUniform A version of WeightInitXavierUniform for sigmoid activation functions.
class	WeightInitUniform Uniform U[-a,a] with a=1/sqrt(fanIn).
class	WeightInitVarScalingNormalFanAvg Truncated aussian distribution with mean 0, variance 1.0/((fanIn + fanOut)/2)
class	WeightInitVarScalingNormalFanIn Gaussian distribution with mean 0, variance 1.0/(fanIn) If a scale is provided, use variance scale/(fanIn) instead
class	WeightInitVarScalingNormalFanOut Truncated normal distribution with mean 0, variance 1.0/(fanOut) If a scale is provided, variance is scale / fanOut
class	WeightInitVarScalingUniformFanAvg Uniform U[-a,a] with a=3.0/((fanIn + fanOut)/2)
class	WeightInitVarScalingUniformFanIn Uniform U[-a,a] with a=3.0/(fanIn) If a scale is provided, a = 3.0 * scale / (fanIn)
class	WeightInitVarScalingUniformFanOut Uniform U[-a,a] with a=3.0/(fanOut) If a scale is provided, a = 3.0 * scale / fanOut
class	WeightInitXavier As per Glorot and Bengio 2010: Gaussian distribution with mean 0, variance 2.0/(fanIn + fanOut)
class	WeightInitXavierLegacy Xavier weight init in DL4J up to 0.6.0.
class	WeightInitXavierUniform As per Glorot and Bengio 2010: Uniform distribution U(-s,s) with s = sqrt(6/(fanIn + fanOut))

Methods in org.deeplearning4j.nn.weights that return IWeightInit

Modifier and Type	Method and Description
IWeightInit	WeightInit.getWeightInitFunction() Create an instance of the weight initialization function
IWeightInit	WeightInit.getWeightInitFunction(Distribution distribution) Create an instance of the weight initialization function

Uses of IWeightInit in org.deeplearning4j.nn.weights.embeddings

Classes in org.deeplearning4j.nn.weights.embeddings that implement IWeightInit

Modifier and Type	Class and Description
class	WeightInitEmbedding Weight initialization for initializing the parameters of an EmbeddingLayer from a EmbeddingInitializer Note: WeightInitEmbedding supports both JSON serializable and non JSON serializable initializations.