org.deeplearning4j.nn.modelimport.keras.layers.recurrent

Class KerasLSTM

java.lang.Object

org.deeplearning4j.nn.modelimport.keras.KerasLayer

org.deeplearning4j.nn.modelimport.keras.layers.recurrent.KerasLSTM

public class KerasLSTM
extends KerasLayer

Imports a Keras LSTM layer as a DL4J LSTM layer.

Author:

dave@skymind.io, Max Pumperla

Nested Class Summary

Nested classes/interfaces inherited from class org.deeplearning4j.nn.modelimport.keras.KerasLayer

KerasLayer.DimOrder

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	returnSequences
protected boolean	unroll

Fields inherited from class org.deeplearning4j.nn.modelimport.keras.KerasLayer

className, conf, dimOrder, dropout, inboundLayerNames, inputShape, kerasMajorVersion, layer, layerName, originalLayerConfig, outboundLayerNames, vertex, weightL1Regularization, weightL2Regularization, weights

Constructor Summary

Constructors

Constructor and Description
KerasLSTM(Integer kerasVersion) Pass-through constructor from KerasLayer
KerasLSTM(Map<String,Object> layerConfig) Constructor from parsed Keras layer configuration dictionary.
KerasLSTM(Map<String,Object> layerConfig, boolean enforceTrainingConfig) Constructor from parsed Keras layer configuration dictionary.
KerasLSTM(Map<String,Object> layerConfig, boolean enforceTrainingConfig, Map<String,? extends KerasLayer> previousLayers) Constructor from parsed Keras layer configuration dictionary.
KerasLSTM(Map<String,Object> layerConfig, Map<String,? extends KerasLayer> previousLayers) Constructor from parsed Keras layer configuration dictionary.

Method Summary

Modifier and Type	Method and Description
double	getForgetBiasInitFromConfig(Map<String,Object> layerConfig, boolean train) Get LSTM forget gate bias initialization from Keras layer configuration.
IActivation	getGateActivationFromConfig(Map<String,Object> layerConfig) Get LSTM gate activation function from Keras layer configuration.
InputPreProcessor	getInputPreprocessor(InputType... inputType) Gets appropriate DL4J InputPreProcessor for given InputTypes.
Layer	getLSTMLayer() Get DL4J Layer.
int	getNumParams() Returns number of trainable parameters in layer.
InputType	getOutputType(InputType... inputType) Get layer output type.
boolean	getUnroll() Get whether LSTM layer should be unrolled (for truncated BPTT).
void	setWeights(Map<String,INDArray> weights) Set weights for layer.

Methods inherited from class org.deeplearning4j.nn.modelimport.keras.KerasLayer

clearCustomLayers, clearLambdaLayers, copyWeightsToLayer, getClassName, getDimOrder, getInboundLayerNames, getInputShape, getKerasMajorVersion, getLayer, getLayerName, getNInFromConfig, getVertex, getWeights, isInputPreProcessor, isLayer, isValidInboundLayer, isVertex, registerCustomLayer, registerLambdaLayer, setInboundLayerNames, setLayer, usesRegularization

Methods inherited from class java.lang.Object

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

Field Detail

unroll

protected boolean unroll

returnSequences

protected boolean returnSequences

Constructor Detail

KerasLSTM

public KerasLSTM(Integer kerasVersion)
          throws UnsupportedKerasConfigurationException

Pass-through constructor from KerasLayer

Parameters:

kerasVersion - major keras version

Throws:

UnsupportedKerasConfigurationException - Unsupported Keras config

KerasLSTM

public KerasLSTM(Map<String,Object> layerConfig)
          throws InvalidKerasConfigurationException,
                 UnsupportedKerasConfigurationException

Constructor from parsed Keras layer configuration dictionary.

Parameters:

layerConfig - dictionary containing Keras layer configuration.

Throws:

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException - Unsupported Keras config

KerasLSTM

public KerasLSTM(Map<String,Object> layerConfig,
                 boolean enforceTrainingConfig)
          throws InvalidKerasConfigurationException,
                 UnsupportedKerasConfigurationException

Constructor from parsed Keras layer configuration dictionary.

Parameters:

layerConfig - dictionary containing Keras layer configuration.

enforceTrainingConfig - whether to load Keras training configuration

Throws:

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException - Unsupported Keras config

KerasLSTM

public KerasLSTM(Map<String,Object> layerConfig,
                 Map<String,? extends KerasLayer> previousLayers)
          throws InvalidKerasConfigurationException,
                 UnsupportedKerasConfigurationException

Constructor from parsed Keras layer configuration dictionary.

Parameters:

layerConfig - dictionary containing Keras layer configuration.

previousLayers - dictionary containing the previous layers in the topology

Throws:

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException - Unsupported Keras config

KerasLSTM

public KerasLSTM(Map<String,Object> layerConfig,
                 boolean enforceTrainingConfig,
                 Map<String,? extends KerasLayer> previousLayers)
          throws InvalidKerasConfigurationException,
                 UnsupportedKerasConfigurationException

Constructor from parsed Keras layer configuration dictionary.

Parameters:

layerConfig - dictionary containing Keras layer configuration

enforceTrainingConfig - whether to enforce training-related configuration options

previousLayers - - dictionary containing the previous layers in the topology

Throws:

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException - Unsupported Keras config

Method Detail

getLSTMLayer

public Layer getLSTMLayer()

Get DL4J Layer. If returnSequences is true, this can be casted to an "LSTM" layer, otherwise it can be casted to a "LastTimeStep" layer.

Returns:

LSTM Layer

getOutputType

public InputType getOutputType(InputType... inputType)
                        throws InvalidKerasConfigurationException

Get layer output type.

Overrides:

getOutputType in class KerasLayer

Parameters:

inputType - Array of InputTypes

Returns:

output type as InputType

Throws:

InvalidKerasConfigurationException - Invalid Keras config

getNumParams

public int getNumParams()

Returns number of trainable parameters in layer.

Overrides:

getNumParams in class KerasLayer

Returns:

number of trainable parameters (12)

getInputPreprocessor

public InputPreProcessor getInputPreprocessor(InputType... inputType)
                                       throws InvalidKerasConfigurationException

Gets appropriate DL4J InputPreProcessor for given InputTypes.

Overrides:

getInputPreprocessor in class KerasLayer

Parameters:

inputType - Array of InputTypes

Returns:

DL4J InputPreProcessor

Throws:

InvalidKerasConfigurationException - Invalid Keras configuration exception

See Also:

InputPreProcessor

setWeights

public void setWeights(Map<String,INDArray> weights)
                throws InvalidKerasConfigurationException

Set weights for layer.

Overrides:

setWeights in class KerasLayer

Parameters:

weights - LSTM layer weights

Throws:

InvalidKerasConfigurationException

getUnroll

public boolean getUnroll()

Get whether LSTM layer should be unrolled (for truncated BPTT).

Returns:

whether to unroll the LSTM

getGateActivationFromConfig

public IActivation getGateActivationFromConfig(Map<String,Object> layerConfig)
                                        throws InvalidKerasConfigurationException,
                                               UnsupportedKerasConfigurationException

Get LSTM gate activation function from Keras layer configuration.

Parameters:

layerConfig - dictionary containing Keras layer configuration

Returns:

LSTM inner activation function

Throws:

InvalidKerasConfigurationException - Invalid Keras config

UnsupportedKerasConfigurationException

getForgetBiasInitFromConfig

public double getForgetBiasInitFromConfig(Map<String,Object> layerConfig,
                                          boolean train)
                                   throws InvalidKerasConfigurationException,
                                          UnsupportedKerasConfigurationException

Get LSTM forget gate bias initialization from Keras layer configuration.

Parameters:

layerConfig - dictionary containing Keras layer configuration

Returns:

LSTM forget gate bias init

Throws:

InvalidKerasConfigurationException - Unsupported Keras config

UnsupportedKerasConfigurationException