org.deeplearning4j.nn.conf.layers.samediff

Class BaseSameDiffLayer

java.lang.Object

org.deeplearning4j.nn.conf.layers.Layer

org.deeplearning4j.nn.conf.layers.samediff.AbstractSameDiffLayer

org.deeplearning4j.nn.conf.layers.samediff.BaseSameDiffLayer

All Implemented Interfaces:

Serializable, Cloneable

public abstract class BaseSameDiffLayer
extends AbstractSameDiffLayer

A base layer used for implementing Deeplearning4j layers using SameDiff. These layers are not scoring/output layers: that is, they should be used as the intermediate layer in a network only. Deeplearning4j SameDiff output layers will be added at a later date.
NOTE: At present, only forward pass is supported. Backward pass will be added at a future date.

To implement a Deeplearinng layer using SameDiff, extend this class.
There are 4 required methods:
- defineLayer: Defines the forward pass for the layer
- defineParameters: Define the layer's parameters in a way suitable for DL4J
- initializeParameters: if required, set the initial parameter values for the layer
- getOutputType: determine the type of output/activations for the layer (without actually executing the layer's forward pass)

Furthermore, there are 3 optional methods:
- setNIn(InputType inputType, boolean override): if implemented, set the number of inputs to the layer during network initialization
- getPreProcessorForInputType: return the preprocessor that should be added (if any), for the given input type
- applyGlobalConfigToLayer: apply any global configuration options (weight init, activation functions etc) to the layer's configuration.

Author:

Alex Black

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	BaseSameDiffLayer.Builder<T extends BaseSameDiffLayer.Builder<T>>

Field Summary

Fields

Modifier and Type	Field and Description
protected WeightInit	weightInit

Fields inherited from class org.deeplearning4j.nn.conf.layers.samediff.AbstractSameDiffLayer

biasUpdater, l1, l1Bias, l2, l2Bias, updater

Fields inherited from class org.deeplearning4j.nn.conf.layers.Layer

constraints, iDropout, layerName

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	BaseSameDiffLayer()
protected	BaseSameDiffLayer(BaseSameDiffLayer.Builder builder)

Method Summary

Modifier and Type	Method and Description
void	applyGlobalConfigToLayer(NeuralNetConfiguration.Builder globalConfig) Apply the global configuration (weight init, activation function, etc) to this layer
abstract List<org.nd4j.autodiff.samediff.SDVariable>	defineLayer(org.nd4j.autodiff.samediff.SameDiff sameDiff, org.nd4j.autodiff.samediff.SDVariable layerInput, Map<String,org.nd4j.autodiff.samediff.SDVariable> paramTable)
InputPreProcessor	getPreProcessorForInputType(InputType inputType) For the given type of input to this layer, what preprocessor (if any) is required? Returns null if no preprocessor is required, otherwise returns an appropriate InputPreProcessor for this layer, such as a CnnToFeedForwardPreProcessor
Layer	instantiate(NeuralNetConfiguration conf, Collection<TrainingListener> trainingListeners, int layerIndex, org.nd4j.linalg.api.ndarray.INDArray layerParamsView, boolean initializeParams)
void	setNIn(InputType inputType, boolean override) Set the nIn value (number of inputs, or input channels for CNNs) based on the given input type

Methods inherited from class org.deeplearning4j.nn.conf.layers.samediff.AbstractSameDiffLayer

applyGlobalConfig, defineParameters, getL1ByParam, getL2ByParam, getLayerParams, getMemoryReport, getOutputType, getUpdaterByParam, initializeParameters, initializer, initWeights, isPretrainParam, paramReshapeOrder

Methods inherited from class org.deeplearning4j.nn.conf.layers.Layer

clone, initializeConstraints, resetLayerDefaultConfig

Methods inherited from class java.lang.Object

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

Field Detail

weightInit

protected WeightInit weightInit

Constructor Detail

BaseSameDiffLayer

protected BaseSameDiffLayer(BaseSameDiffLayer.Builder builder)

BaseSameDiffLayer

protected BaseSameDiffLayer()

Method Detail

defineLayer

public abstract List<org.nd4j.autodiff.samediff.SDVariable> defineLayer(org.nd4j.autodiff.samediff.SameDiff sameDiff,
                                                                        org.nd4j.autodiff.samediff.SDVariable layerInput,
                                                                        Map<String,org.nd4j.autodiff.samediff.SDVariable> paramTable)

setNIn

public void setNIn(InputType inputType,
                   boolean override)

Description copied from class: Layer

Set the nIn value (number of inputs, or input channels for CNNs) based on the given input type

Specified by:

setNIn in class AbstractSameDiffLayer

Parameters:

inputType - Input type for this layer

override - If false: only set the nIn value if it's not already set. If true: set it regardless of whether it's already set or not.

getPreProcessorForInputType

public InputPreProcessor getPreProcessorForInputType(InputType inputType)

Description copied from class: Layer

For the given type of input to this layer, what preprocessor (if any) is required?
Returns null if no preprocessor is required, otherwise returns an appropriate InputPreProcessor for this layer, such as a CnnToFeedForwardPreProcessor

Specified by:

getPreProcessorForInputType in class AbstractSameDiffLayer

Parameters:

inputType - InputType to this layer

Returns:

Null if no preprocessor is required, otherwise the type of preprocessor necessary for this layer/input combination

applyGlobalConfigToLayer

public void applyGlobalConfigToLayer(NeuralNetConfiguration.Builder globalConfig)

Description copied from class: AbstractSameDiffLayer

Apply the global configuration (weight init, activation function, etc) to this layer

Specified by:

applyGlobalConfigToLayer in class AbstractSameDiffLayer

Parameters:

globalConfig - Global configuration

instantiate

public Layer instantiate(NeuralNetConfiguration conf,
                         Collection<TrainingListener> trainingListeners,
                         int layerIndex,
                         org.nd4j.linalg.api.ndarray.INDArray layerParamsView,
                         boolean initializeParams)

Specified by:

instantiate in class AbstractSameDiffLayer