ConcatenatedModel concatenates two models such that the output of the first model is input to the second. More...

#include <shark/Models/ConcatenatedModel.h>

Inheritance diagram for shark::ConcatenatedModel< VectorType >:

Public Types
typedef base_type::BatchInputType	BatchInputType

typedef base_type::BatchOutputType	BatchOutputType

typedef base_type::ParameterVectorType	ParameterVectorType

Public Types inherited from shark::AbstractModel< VectorType, VectorType, VectorType >
enum	Feature

typedef VectorType	InputType
	Defines the input type of the model. More...

typedef VectorType	OutputType
	Defines the output type of the model. More...

typedef AbstractModel< VectorType, VectorType, VectorType >	ModelBaseType
	Defines the BaseType used by the model (this type). Useful for creating derived models. More...

typedef Batch< InputType >::type	BatchInputType
	defines the batch type of the input type. More...

typedef Batch< OutputType >::type	BatchOutputType
	defines the batch type of the output type More...

typedef TypedFlags< Feature >	Features

typedef TypedFeatureNotAvailableException< Feature >	FeatureNotAvailableException

Public Types inherited from shark::IParameterizable< VectorType >
typedef VectorType	ParameterVectorType

Public Member Functions
std::string	name () const
	From INameable: return the class name. More...

Shape	inputShape () const
	Returns the expected shape of the input. More...

Shape	outputShape () const
	Returns the shape of the output. More...

void	add (AbstractModel< VectorType, VectorType > *layer, bool optimize)

void	enableModelOptimization (std::size_t index, bool opt)
	sets whether the parameters of the index-th model should be optimized More...

ParameterVectorType	parameterVector () const
	Return the parameter vector. More...

void	setParameterVector (ParameterVectorType const &newParameters)
	Set the parameter vector. More...

std::size_t	numberOfParameters () const
	Return the number of parameters. More...

boost::shared_ptr< State >	createState () const
	Creates an internal state of the model. More...

BatchOutputType const &	hiddenResponses (State const &state, std::size_t index) const

State const &	hiddenState (State const &state, std::size_t index) const

void	eval (BatchInputType const &patterns, BatchOutputType &outputs) const
	Standard interface for evaluating the response of the model to a batch of patterns. More...

void	eval (BatchInputType const &patterns, BatchOutputType &outputs, State &state) const
	Standard interface for evaluating the response of the model to a batch of patterns. More...

void	weightedParameterDerivative (BatchInputType const &patterns, BatchOutputType const &outputs, BatchOutputType const &coefficients, State const &state, RealVector &gradient) const
	calculates the weighted sum of derivatives w.r.t the parameters. More...

void	weightedInputDerivative (BatchInputType const &patterns, BatchOutputType const &outputs, BatchOutputType const &coefficients, State const &state, BatchOutputType &derivatives) const

virtual void	weightedDerivatives (BatchInputType const &patterns, BatchOutputType const &outputs, BatchOutputType const &coefficients, State const &state, RealVector &gradient, BatchInputType &inputDerivative) const
	calculates weighted input and parameter derivative at the same time More...

void	read (InArchive &archive)
	From ISerializable. More...

void	write (OutArchive &archive) const
	From ISerializable. More...

Public Member Functions inherited from shark::AbstractModel< VectorType, VectorType, VectorType >
	AbstractModel ()

virtual	~AbstractModel ()

const Features &	features () const

virtual void	updateFeatures ()

bool	hasFirstParameterDerivative () const
	Returns true when the first parameter derivative is implemented. More...

bool	hasFirstInputDerivative () const
	Returns true when the first input derivative is implemented. More...

virtual void	eval (InputType const &pattern, OutputType &output) const
	Standard interface for evaluating the response of the model to a single pattern. More...

Data< OutputType >	operator() (Data< InputType > const &patterns) const
	Model evaluation as an operator for a whole dataset. This is a convenience function. More...

OutputType	operator() (InputType const &pattern) const
	Model evaluation as an operator for a single pattern. This is a convenience function. More...

BatchOutputType	operator() (BatchInputType const &patterns) const
	Model evaluation as an operator for a single pattern. This is a convenience function. More...

virtual void	weightedInputDerivative (BatchInputType const &pattern, BatchOutputType const &outputs, BatchOutputType const &coefficients, State const &state, BatchInputType &derivative) const
	calculates the weighted sum of derivatives w.r.t the inputs More...

Public Member Functions inherited from shark::IParameterizable< VectorType >
virtual	~IParameterizable ()

Public Member Functions inherited from shark::INameable
virtual	~INameable ()

Public Member Functions inherited from shark::ISerializable
virtual	~ISerializable ()
	Virtual d'tor. More...

void	load (InArchive &archive, unsigned int version)
	Versioned loading of components, calls read(...). More...

void	save (OutArchive &archive, unsigned int version) const
	Versioned storing of components, calls write(...). More...

	BOOST_SERIALIZATION_SPLIT_MEMBER ()

Additional Inherited Members
Protected Attributes inherited from shark::AbstractModel< VectorType, VectorType, VectorType >
Features	m_features

Detailed Description

template<class VectorType>
class shark::ConcatenatedModel< VectorType >

ConcatenatedModel concatenates two models such that the output of the first model is input to the second.

Sometimes a series of models is needed to generate the desired output. For example when input data needs to be normalized before it can be put into the trained model. In this case, the ConcatenatedModel can be used to represent this series as one model. The easiest way to do is is using the operator >> of AbstractModel: ConcatenatedModel<InputType,OutputType> model = model1>>model2; InputType must be the type of input model1 receives and model2 the output of model2. The output of model1 and input of model2 must match. Another way of construction is calling the constructor of ConcatenatedModel using the constructor: ConcatenatedModel<InputType,OutputType> model (&modell,&model2); warning: model1 and model2 must outlive model. When they are destroyed first, behavior is undefined.

Definition at line 56 of file ConcatenatedModel.h.

Member Typedef Documentation

◆ BatchInputType

template<class VectorType>

typedef base_type::BatchInputType shark::ConcatenatedModel< VectorType >::BatchInputType

Definition at line 60 of file ConcatenatedModel.h.

◆ BatchOutputType

template<class VectorType>

typedef base_type::BatchOutputType shark::ConcatenatedModel< VectorType >::BatchOutputType

Definition at line 61 of file ConcatenatedModel.h.

◆ ParameterVectorType

template<class VectorType>

typedef base_type::ParameterVectorType shark::ConcatenatedModel< VectorType >::ParameterVectorType

Definition at line 62 of file ConcatenatedModel.h.

Member Function Documentation

◆ add()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::add	(	AbstractModel< VectorType, VectorType > *	layer,
		bool	optimize
	)

inline

Definition at line 79 of file ConcatenatedModel.h.

References shark::ConcatenatedModel< VectorType >::enableModelOptimization().

Referenced by shark::operator>>().

◆ createState()

template<class VectorType>

boost::shared_ptr<State> shark::ConcatenatedModel< VectorType >::createState ( ) const

inlinevirtual

Creates an internal state of the model.

The state is needed when the derivatives are to be calculated. Eval can store a state which is then reused to speed up the calculations of the derivatives. This also allows eval to be evaluated in parallel!

Reimplemented from shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 145 of file ConcatenatedModel.h.

◆ enableModelOptimization()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::enableModelOptimization	(	std::size_t	index,
		bool	opt
	)

inline

sets whether the parameters of the index-th model should be optimized

If the model has non-differentiable submodels disabling those will make the whole model differentiable. Note that the models are ordered as model0 >> model1>> model2>>...

Definition at line 89 of file ConcatenatedModel.h.

References shark::AbstractModel< VectorType, VectorType, VectorType >::HAS_FIRST_INPUT_DERIVATIVE, shark::AbstractModel< VectorType, VectorType, VectorType >::HAS_FIRST_PARAMETER_DERIVATIVE, shark::AbstractModel< VectorType, VectorType, VectorType >::m_features, shark::TypedFlags< Flag >::reset(), and SIZE_CHECK.

Referenced by shark::ConcatenatedModel< VectorType >::add(), and trainAutoencoderModel().

◆ eval() [1/2]

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::eval	(	BatchInputType const &	patterns,
		BatchOutputType &	outputs
	)		const

inlinevirtual

Standard interface for evaluating the response of the model to a batch of patterns.

Parameters

patterns	the inputs of the model
outputs	the predictions or response of the model to every pattern

Reimplemented from shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 165 of file ConcatenatedModel.h.

References shark::swap().

◆ eval() [2/2]

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::eval	(	BatchInputType const &	patterns,
		BatchOutputType &	outputs,
		State &	state
	)		const

inlinevirtual

Standard interface for evaluating the response of the model to a batch of patterns.

Parameters

patterns	the inputs of the model
outputs	the predictions or response of the model to every pattern
state	intermediate results stored by eval which can be reused for derivative computation.

Implements shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 173 of file ConcatenatedModel.h.

References shark::State::toState().

◆ hiddenResponses()

template<class VectorType>

BatchOutputType const& shark::ConcatenatedModel< VectorType >::hiddenResponses	(	State const &	state,
		std::size_t	index
	)		const

inline

Definition at line 154 of file ConcatenatedModel.h.

References shark::State::toState().

◆ hiddenState()

template<class VectorType>

State const& shark::ConcatenatedModel< VectorType >::hiddenState	(	State const &	state,
		std::size_t	index
	)		const

inline

Definition at line 159 of file ConcatenatedModel.h.

References shark::AbstractModel< VectorType, VectorType, VectorType >::eval(), and shark::State::toState().

◆ inputShape()

template<class VectorType>

Shape shark::ConcatenatedModel< VectorType >::inputShape ( ) const

inlinevirtual

Returns the expected shape of the input.

Implements shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 70 of file ConcatenatedModel.h.

◆ name()

template<class VectorType>

std::string shark::ConcatenatedModel< VectorType >::name ( ) const

inlinevirtual

From INameable: return the class name.

Reimplemented from shark::INameable.

Definition at line 65 of file ConcatenatedModel.h.

Referenced by trainAutoencoderModel().

◆ numberOfParameters()

template<class VectorType>

std::size_t shark::ConcatenatedModel< VectorType >::numberOfParameters ( ) const

inlinevirtual

Return the number of parameters.

Reimplemented from shark::IParameterizable< VectorType >.

Definition at line 136 of file ConcatenatedModel.h.

Referenced by shark::ConcatenatedModel< VectorType >::parameterVector(), shark::ConcatenatedModel< VectorType >::weightedDerivatives(), and shark::ConcatenatedModel< VectorType >::weightedParameterDerivative().

◆ outputShape()

template<class VectorType>

Shape shark::ConcatenatedModel< VectorType >::outputShape ( ) const

inlinevirtual

Returns the shape of the output.

Implements shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 74 of file ConcatenatedModel.h.

◆ parameterVector()

template<class VectorType>

ParameterVectorType shark::ConcatenatedModel< VectorType >::parameterVector ( ) const

inlinevirtual

Return the parameter vector.

Reimplemented from shark::IParameterizable< VectorType >.

Definition at line 114 of file ConcatenatedModel.h.

References shark::ConcatenatedModel< VectorType >::numberOfParameters().

◆ read()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::read ( InArchive & archive )

inlinevirtual

From ISerializable.

Reimplemented from shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 277 of file ConcatenatedModel.h.

◆ setParameterVector()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::setParameterVector ( ParameterVectorType const & newParameters )

inlinevirtual

Set the parameter vector.

Reimplemented from shark::IParameterizable< VectorType >.

Definition at line 126 of file ConcatenatedModel.h.

Referenced by trainAutoencoderModel().

◆ weightedDerivatives()

template<class VectorType>

virtual void shark::ConcatenatedModel< VectorType >::weightedDerivatives	(	BatchInputType const &	patterns,
		BatchOutputType const &	outputs,
		BatchOutputType const &	coefficients,
		State const &	state,
		RealVector &	parameterDerivative,
		BatchInputType &	inputDerivative
	)		const

inlinevirtual

calculates weighted input and parameter derivative at the same time

Sometimes, both derivatives are needed at the same time. But sometimes, when calculating the weighted parameter derivative, the input derivative can be calculated for free. This is for example true for the feed-forward neural networks. However, there exists the obvious default implementation to just calculate the derivatives one after another.

Parameters

patterns	the patterns to evaluate
coefficients	the coefficients which are used to calculate the weighted sum
state	intermediate results stored by eval to sped up calculations of the derivatives
parameterDerivative	the calculated parameter derivative as sum over all derivates of all patterns
inputDerivative	the calculated derivative for every pattern

Reimplemented from shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 244 of file ConcatenatedModel.h.

References shark::ConcatenatedModel< VectorType >::numberOfParameters(), shark::swap(), and shark::State::toState().

◆ weightedInputDerivative()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::weightedInputDerivative	(	BatchInputType const &	patterns,
		BatchOutputType const &	outputs,
		BatchOutputType const &	coefficients,
		State const &	state,
		BatchOutputType &	derivatives
	)		const

inline

Definition at line 222 of file ConcatenatedModel.h.

References shark::swap(), and shark::State::toState().

◆ weightedParameterDerivative()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::weightedParameterDerivative	(	BatchInputType const &	pattern,
		BatchOutputType const &	outputs,
		BatchOutputType const &	coefficients,
		State const &	state,
		RealVector &	derivative
	)		const

inlinevirtual

calculates the weighted sum of derivatives w.r.t the parameters.

Parameters

pattern	the patterns to evaluate
coefficients	the coefficients which are used to calculate the weighted sum for every pattern
state	intermediate results stored by eval to speed up calculations of the derivatives
derivative	the calculated derivative as sum over all derivates of all patterns

Reimplemented from shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 185 of file ConcatenatedModel.h.

References shark::ConcatenatedModel< VectorType >::numberOfParameters(), shark::swap(), and shark::State::toState().

◆ write()

template<class VectorType>

void shark::ConcatenatedModel< VectorType >::write ( OutArchive & archive ) const

inlinevirtual

From ISerializable.

Reimplemented from shark::AbstractModel< VectorType, VectorType, VectorType >.

Definition at line 285 of file ConcatenatedModel.h.

The documentation for this class was generated from the following file:

include/shark/Models/ConcatenatedModel.h

Public Types

Public Member Functions

Additional Inherited Members

Detailed Description

template<class VectorType> class shark::ConcatenatedModel< VectorType >

Member Typedef Documentation

◆ BatchInputType

◆ BatchOutputType

◆ ParameterVectorType

Member Function Documentation

◆ add()

◆ createState()

◆ enableModelOptimization()

◆ eval() [1/2]

◆ eval() [2/2]

◆ hiddenResponses()

◆ hiddenState()

◆ inputShape()

◆ name()

◆ numberOfParameters()

◆ outputShape()

◆ parameterVector()

◆ read()

◆ setParameterVector()

◆ weightedDerivatives()

◆ weightedInputDerivative()

◆ weightedParameterDerivative()

◆ write()

template<class VectorType>
class shark::ConcatenatedModel< VectorType >