ldaplusplus::em::AbstractEStep< Scalar > Class Template Reference

#include <AbstractEStep.hpp>

Inheritance diagram for ldaplusplus::em::AbstractEStep< Scalar >:

Public Member Functions
	AbstractEStep (int random_state)
virtual void	e_step () override
Public Member Functions inherited from ldaplusplus::em::EStepInterface< Scalar >
virtual std::shared_ptr< parameters::Parameters >	doc_e_step (const std::shared_ptr< corpus::Document > doc, const std::shared_ptr< parameters::Parameters > parameters)=0
Public Member Functions inherited from ldaplusplus::events::EventDispatcherComposition
std::shared_ptr< EventDispatcherInterface >	get_event_dispatcher ()
void	set_event_dispatcher (std::shared_ptr< EventDispatcherInterface > dispatcher)

Protected Member Functions
bool	converged (const Eigen::Matrix< Scalar, Eigen::Dynamic, 1 > &gamma_old, const Eigen::Matrix< Scalar, Eigen::Dynamic, 1 > &gamma, Scalar tolerance)
PRNG &	get_prng ()

Detailed Description

template<typename Scalar>
class ldaplusplus::em::AbstractEStep< Scalar >

A base class that provides few common functionalities for implementing an E step.

• Implements an empty void e_step() since most work happens in doc_e_step()
• Provides convergence check based on variational parameter \(\gamma\)
• Provides a PRNG initialized using a seed in the constructor

Constructor & Destructor Documentation

template<typename Scalar >

ldaplusplus::em::AbstractEStep< Scalar >::AbstractEStep

(

int

random_state

)

Require a random state to be passed from the extending classes.

Parameters

random_state

An initial seed value for random number generation.

Member Function Documentation

template<typename Scalar >

bool ldaplusplus::em::AbstractEStep< Scalar >::converged ( const Eigen::Matrix< Scalar, Eigen::Dynamic, 1 > &  gamma_old, const Eigen::Matrix< Scalar, Eigen::Dynamic, 1 > &  gamma, Scalar  tolerance  )

protected

Check for convergence based on the mean relative change of the variational parameter \(\gamma\).

Parameters

gamma_old	The gamma of the previous iteration.
gamma	The gamma of this iteration.
tolerance	The threshold below which we declare convergence.

Returns

Whether the change is small enough to indicate convergence.

template<typename Scalar >

virtual void ldaplusplus::em::AbstractEStep< Scalar >::e_step ( )

inlineoverridevirtual

Implement an empty e_step because almost nobody needs to perform some action at the end of each corpus epoch.

Implements ldaplusplus::em::EStepInterface< Scalar >.

Reimplemented in ldaplusplus::em::FastSupervisedEStep< Scalar >.

template<typename Scalar >

PRNG& ldaplusplus::em::AbstractEStep< Scalar >::get_prng ( )

inlineprotected

Return a PRNG for use with any distribution.

Although this isn't all that different from making random_ protected it could allow for future change of the object returned since it can be anything that satisfies the UniformRandomBitGenerator (see: http://en.cppreference.com/w/cpp/concept/UniformRandomBitGenerator).

---

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

• include/ldaplusplus/em/AbstractEStep.hpp
• src/ldaplusplus/em/AbstractEStep.cpp