org.deeplearning4j.nn.params

Class DefaultParamInitializer

java.lang.Object

org.deeplearning4j.nn.params.DefaultParamInitializer

All Implemented Interfaces:

ParamInitializer

Direct Known Subclasses:

CenterLossParamInitializer, PretrainParamInitializer, VariationalAutoencoderParamInitializer

public class DefaultParamInitializer
extends java.lang.Object
implements ParamInitializer

Static weight initializer with just a weight matrix and a bias

Field Summary

Fields

Modifier and Type	Field and Description
static java.lang.String	BIAS_KEY
static java.lang.String	WEIGHT_KEY

Constructor Summary

Constructors

Constructor and Description
DefaultParamInitializer()

Method Summary

Modifier and Type	Method and Description
protected org.nd4j.linalg.api.ndarray.INDArray	createBias(int nOut, double biasInit, org.nd4j.linalg.api.ndarray.INDArray biasParamView, boolean initializeParameters)
protected org.nd4j.linalg.api.ndarray.INDArray	createBias(NeuralNetConfiguration conf, org.nd4j.linalg.api.ndarray.INDArray biasParamView, boolean initializeParameters)
protected org.nd4j.linalg.api.ndarray.INDArray	createWeightMatrix(int nIn, int nOut, WeightInit weightInit, org.nd4j.linalg.api.rng.distribution.Distribution dist, org.nd4j.linalg.api.ndarray.INDArray weightParamView, boolean initializeParameters)
protected org.nd4j.linalg.api.ndarray.INDArray	createWeightMatrix(NeuralNetConfiguration conf, org.nd4j.linalg.api.ndarray.INDArray weightParamView, boolean initializeParameters)
java.util.Map<java.lang.String,org.nd4j.linalg.api.ndarray.INDArray>	getGradientsFromFlattened(NeuralNetConfiguration conf, org.nd4j.linalg.api.ndarray.INDArray gradientView) Return a map of gradients (in their standard non-flattened representation), taken from the flattened (row vector) gradientView array.
static DefaultParamInitializer	getInstance()
java.util.Map<java.lang.String,org.nd4j.linalg.api.ndarray.INDArray>	init(NeuralNetConfiguration conf, org.nd4j.linalg.api.ndarray.INDArray paramsView, boolean initializeParams) Initialize the parameters
int	numParams(NeuralNetConfiguration conf)

Methods inherited from class java.lang.Object

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

Field Detail

WEIGHT_KEY

public static final java.lang.String WEIGHT_KEY

See Also:

Constant Field Values

BIAS_KEY

public static final java.lang.String BIAS_KEY

See Also:

Constant Field Values

Constructor Detail

DefaultParamInitializer

public DefaultParamInitializer()

Method Detail

getInstance

public static DefaultParamInitializer getInstance()

numParams

public int numParams(NeuralNetConfiguration conf)

Specified by:

numParams in interface ParamInitializer

init

public java.util.Map<java.lang.String,org.nd4j.linalg.api.ndarray.INDArray> init(NeuralNetConfiguration conf,
                                                                                 org.nd4j.linalg.api.ndarray.INDArray paramsView,
                                                                                 boolean initializeParams)

Description copied from interface: ParamInitializer

Initialize the parameters

Specified by:

init in interface ParamInitializer

Parameters:

conf - the configuration

paramsView - a view of the full network (backprop) parameters

initializeParams - if true: initialize the parameters according to the configuration. If false: don't modify the values in the paramsView array (but do select out the appropriate subset, reshape etc as required)

Returns:

Map of parameters keyed by type (view of the 'paramsView' array)

getGradientsFromFlattened

public java.util.Map<java.lang.String,org.nd4j.linalg.api.ndarray.INDArray> getGradientsFromFlattened(NeuralNetConfiguration conf,
                                                                                                      org.nd4j.linalg.api.ndarray.INDArray gradientView)

Description copied from interface: ParamInitializer

Return a map of gradients (in their standard non-flattened representation), taken from the flattened (row vector) gradientView array. The idea is that operates in exactly the same way as the the paramsView does in #init(Map, NeuralNetConfiguration, INDArray); thus the position in the view (and, the array orders) must match those of the parameters

Specified by:

getGradientsFromFlattened in interface ParamInitializer

Parameters:

conf - Configuration

gradientView - The flattened gradients array, as a view of the larger array

Returns:

A map containing an array by parameter type, that is a view of the full network gradients array

createBias

protected org.nd4j.linalg.api.ndarray.INDArray createBias(NeuralNetConfiguration conf,
                                                          org.nd4j.linalg.api.ndarray.INDArray biasParamView,
                                                          boolean initializeParameters)

createBias

protected org.nd4j.linalg.api.ndarray.INDArray createBias(int nOut,
                                                          double biasInit,
                                                          org.nd4j.linalg.api.ndarray.INDArray biasParamView,
                                                          boolean initializeParameters)

createWeightMatrix

protected org.nd4j.linalg.api.ndarray.INDArray createWeightMatrix(NeuralNetConfiguration conf,
                                                                  org.nd4j.linalg.api.ndarray.INDArray weightParamView,
                                                                  boolean initializeParameters)

createWeightMatrix

protected org.nd4j.linalg.api.ndarray.INDArray createWeightMatrix(int nIn,
                                                                  int nOut,
                                                                  WeightInit weightInit,
                                                                  org.nd4j.linalg.api.rng.distribution.Distribution dist,
                                                                  org.nd4j.linalg.api.ndarray.INDArray weightParamView,
                                                                  boolean initializeParameters)