IT++ Logo
Public Member Functions | Protected Member Functions | Protected Attributes | List of all members
itpp::MOG_generic Class Reference

Generic Mixture of Gaussians (MOG) class. Used as a base for other MOG classes. More...

#include <itpp/stat/mog_generic.h>

Inheritance diagram for itpp::MOG_generic:
itpp::MOG_diag itpp::MOG_diag_EM_sup itpp::MOG_diag_kmeans_sup

Public Member Functions

 MOG_generic ()
 Default constructor.
 
 MOG_generic (const std::string &name_in)
 Construct the MOG_generic object by loading the parameters from a model file.
 
 MOG_generic (const int &K_in, const int &D_in, bool full_in=false)
 construct a default model (all Gaussians have zero mean and unit variance for all dimensions)
 
 MOG_generic (Array< vec > &means_in, bool full_in=false)
 Construct a model using user supplied mean vectors.
 
 MOG_generic (Array< vec > &means_in, Array< vec > &diag_covs_in, vec &weights_in)
 Construct a model using user supplied parameters (diagonal covariance version)
 
 MOG_generic (Array< vec > &means_in, Array< mat > &full_covs_in, vec &weights_in)
 Construct a model using user supplied parameters (full covariance version)
 
virtual ~MOG_generic ()
 Default destructor.
 
void init ()
 Initialise the model to be empty.
 
void init (const int &K_in, const int &D_in, bool full_in=false)
 initialise the model so that all Gaussians have zero mean and unit variance for all dimensions
 
void init (Array< vec > &means_in, bool full_in=false)
 Initialise the model using user supplied mean vectors.
 
void init (Array< vec > &means_in, Array< vec > &diag_covs_in, vec &weights_in)
 Initialise the model using user supplied parameters (diagonal covariance version)
 
void init (Array< vec > &means_in, Array< mat > &full_covs_in, vec &weights_in)
 Initialise the model using user supplied parameters (full covariance version)
 
virtual void cleanup ()
 Release memory used by the model. The model will be empty.
 
bool is_valid () const
 Returns true if the model's parameters are valid.
 
bool is_full () const
 Returns true if the model has full covariance matrices.
 
int get_K () const
 Return the number of Gaussians.
 
int get_D () const
 Return the dimensionality.
 
vec get_weights () const
 Obtain a copy of the weight vector.
 
Array< vec > get_means () const
 Obtain a copy of the array of mean vectors.
 
Array< vec > get_diag_covs () const
 Obtain a copy of the array of diagonal covariance vectors.
 
Array< mat > get_full_covs () const
 Obtain a copy of the array of full covariance matrices.
 
void set_means (Array< vec > &means_in)
 Set the means of the model.
 
void set_diag_covs (Array< vec > &diag_covs_in)
 Set the diagonal covariance vectors of the model.
 
void set_full_covs (Array< mat > &full_covs_in)
 Set the full covariance matrices of the model.
 
void set_weights (vec &weights_in)
 Set the weight vector of the model.
 
void set_means_zero ()
 Set the means in the model to be zero.
 
void set_diag_covs_unity ()
 Set the diagonal covariance vectors to be unity.
 
void set_full_covs_unity ()
 Set the full covariance matrices to be unity.
 
void set_weights_uniform ()
 Set all the weights to 1/K, where K is the number of Gaussians.
 
void set_checks (bool do_checks_in)
 Enable/disable internal checks for likelihood functions.
 
void set_paranoid (bool paranoid_in)
 Enable/disable paranoia about numerical stability.
 
virtual void load (const std::string &name_in)
 Initialise the model by loading the parameters from a model file.
 
virtual void save (const std::string &name_in) const
 Save the model's parameters to a model file.
 
virtual void join (const MOG_generic &B_in)
 Mathematically join the model with a user supplied model.
 
virtual void convert_to_diag ()
 Convert the model to use diagonal covariances.
 
virtual void convert_to_full ()
 Convert the model to have full covariance matrices.
 
virtual double log_lhood_single_gaus (const vec &x_in, const int k)
 calculate the log likelihood of vector x_in using only Gaussian k
 
virtual double log_lhood (const vec &x_in)
 calculate the log likelihood of vector x_in
 
virtual double lhood (const vec &x_in)
 calculate the likelihood of vector x_in
 
virtual double avg_log_lhood (const Array< vec > &X_in)
 calculate the average log likelihood of an array of vectors X_in
 

Protected Member Functions

bool check_size (const vec &x_in) const
 Check if vector x_in has the same dimensionality as the model.
 
bool check_size (const Array< vec > &X_in) const
 Check if all vectors in Array X_in have the same dimensionality as the model.
 
bool check_array_uniformity (const Array< vec > &A) const
 Check if all vectors in Array X_in have the same dimensionality.
 
void set_means_internal (Array< vec > &means_in)
 ADD DOCUMENTATION HERE.
 
void set_diag_covs_internal (Array< vec > &diag_covs_in)
 ADD DOCUMENTATION HERE.
 
void set_full_covs_internal (Array< mat > &full_covs_in)
 ADD DOCUMENTATION HERE.
 
void set_weights_internal (vec &_weigths)
 ADD DOCUMENTATION HERE.
 
void set_means_zero_internal ()
 ADD DOCUMENTATION HERE.
 
void set_diag_covs_unity_internal ()
 ADD DOCUMENTATION HERE.
 
void set_full_covs_unity_internal ()
 ADD DOCUMENTATION HERE.
 
void set_weights_uniform_internal ()
 ADD DOCUMENTATION HERE.
 
void convert_to_diag_internal ()
 ADD DOCUMENTATION HERE.
 
void convert_to_full_internal ()
 ADD DOCUMENTATION HERE.
 
virtual void setup_means ()
 additional processing of mean vectors, done as the last step of mean initialisation
 
virtual void setup_covs ()
 additional processing of covariance vectors/matrices, done as the last step of covariance initialisation
 
virtual void setup_weights ()
 additional processing of the weight vector, done as the last step of weight initialisation
 
virtual void setup_misc ()
 additional processing of miscellaneous parameters, done as the last step of overall initialisation
 
virtual double log_lhood_single_gaus_internal (const vec &x_in, const int k)
 ADD DOCUMENTATION HERE.
 
virtual double log_lhood_internal (const vec &x_in)
 ADD DOCUMENTATION HERE.
 
virtual double lhood_internal (const vec &x_in)
 ADD DOCUMENTATION HERE.
 

Protected Attributes

bool do_checks
 indicates whether checks on input data are done
 
bool valid
 indicates whether the parameters are valid
 
bool full
 indicates whether we are using full or diagonal covariance matrices
 
bool paranoid
 indicates whether we are paranoid about numerical stability
 
int K
 number of gaussians
 
int D
 dimensionality
 
Array< vec > means
 means
 
Array< vec > diag_covs
 diagonal covariance matrices, stored as vectors
 
Array< mat > full_covs
 full covariance matrices
 
vec weights
 weights
 
double log_max_K
 Pre-calcualted std::log(std::numeric_limits<double>::max() / K), where K is the number of Gaussians.
 
vec log_det_etc
 Gaussian specific pre-calcualted constants.
 
vec log_weights
 Pre-calculated log versions of the weights.
 
Array< mat > full_covs_inv
 Pre-calcuated inverted version of each full covariance matrix.
 
Array< vec > diag_covs_inv_etc
 Pre-calcuated inverted version of each diagonal covariance vector, where the covariance elements are first multiplied by two.
 

Detailed Description

Generic Mixture of Gaussians (MOG) class. Used as a base for other MOG classes.

Author
Conrad Sanderson

Used for representing a statistical distribution as a convex combination of multi-variate Gaussian functions. Also known as a Gaussian Mixture Model. This class handles both full and diagonal covariance matrices, allows loading and saving of the MOG's parameters, as well as calculation of likelihoods. The parameters are set by the user or an optimisation algorithm (for example, see the MOG_diag_EM class). For speed and space reasons, diagonal and full covariance matrices are stored and handled separately.

Definition at line 57 of file mog_generic.h.

Constructor & Destructor Documentation

itpp::MOG_generic::MOG_generic ( )
inline

Default constructor.

Note
An empty model is created. The likelihood functions are not useable until the model's parameters are set

Definition at line 67 of file mog_generic.h.

itpp::MOG_generic::MOG_generic ( const std::string &  name_in)
inline

Construct the MOG_generic object by loading the parameters from a model file.

Parameters
name_inThe model's filename

Definition at line 72 of file mog_generic.h.

itpp::MOG_generic::MOG_generic ( const int &  K_in,
const int &  D_in,
bool  full_in = false 
)
inline

construct a default model (all Gaussians have zero mean and unit variance for all dimensions)

Parameters
K_inNumber of Gaussians
D_inDimensionality
full_inIf true, use full covariance matrices; if false, use diagonal covariance matrices. Default = false.

Definition at line 79 of file mog_generic.h.

itpp::MOG_generic::MOG_generic ( Array< vec > &  means_in,
bool  full_in = false 
)
inline

Construct a model using user supplied mean vectors.

Parameters
means_inArray of mean vectors
full_inIf true, use full covariance matrices; if false, use diagonal covariance matrices. Default = false.
Note
The number of mean vectors specifies the number of Gaussians. The covariance matrices are set to the identity matrix. The weights for all Gaussians are the same, equal to 1/K, where K is the number of Gaussians

Definition at line 88 of file mog_generic.h.

itpp::MOG_generic::MOG_generic ( Array< vec > &  means_in,
Array< vec > &  diag_covs_in,
vec &  weights_in 
)
inline

Construct a model using user supplied parameters (diagonal covariance version)

Parameters
means_inArray of mean vectors
diag_covs_inArray of vectors representing diagonal covariances
weights_invector of weights
Note
The number of mean vectors, covariance vectors and weights must be the same

Definition at line 96 of file mog_generic.h.

itpp::MOG_generic::MOG_generic ( Array< vec > &  means_in,
Array< mat > &  full_covs_in,
vec &  weights_in 
)
inline

Construct a model using user supplied parameters (full covariance version)

Parameters
means_inArray of mean vectors
full_covs_inArray of full covariance matrices
weights_invector of weights
Note
The number of mean vectors, covariance matrices and weights must be the same

Definition at line 104 of file mog_generic.h.

Member Function Documentation

void itpp::MOG_generic::init ( )

Initialise the model to be empty.

Note
The likelihood functions are not useable until the model's parameters are set

Definition at line 43 of file mog_generic.cpp.

References cleanup().

Referenced by join(), load(), itpp::MOG_diag_EM_sup::ml(), and itpp::MOG_diag_kmeans_sup::run().

void itpp::MOG_generic::init ( const int &  K_in,
const int &  D_in,
bool  full_in = false 
)

initialise the model so that all Gaussians have zero mean and unit variance for all dimensions

Parameters
K_inNumber of Gaussians
D_inDimensionality
full_inIf true, use full covariance matrices; if false, use diagonal covariance matrices. Default = false.

Definition at line 46 of file mog_generic.cpp.

References D, do_checks, full, it_assert, K, paranoid, set_diag_covs_unity_internal(), set_full_covs_unity_internal(), set_means_zero_internal(), set_weights_uniform_internal(), setup_misc(), and valid.

void itpp::MOG_generic::init ( Array< vec > &  means_in,
bool  full_in = false 
)

Initialise the model using user supplied mean vectors.

Parameters
means_inArray of mean vectors
full_inIf true, use full covariance matrices; if false, use diagonal covariance matrices. Default = false.
Note
The number of mean vectors specifies the number of Gaussians. The covariance matrices are set to the identity matrix. The weights for all Gaussians are the same, equal to 1/K, where K is the number of Gaussians

Definition at line 69 of file mog_generic.cpp.

References check_array_uniformity(), D, do_checks, full, it_assert, K, paranoid, set_diag_covs_unity_internal(), set_full_covs_unity_internal(), set_means(), set_weights_uniform_internal(), setup_misc(), itpp::Array< T >::size(), and valid.

void itpp::MOG_generic::init ( Array< vec > &  means_in,
Array< vec > &  diag_covs_in,
vec &  weights_in 
)

Initialise the model using user supplied parameters (diagonal covariance version)

Parameters
means_inArray of mean vectors
diag_covs_inArray of vectors representing diagonal covariances
weights_invector of weights
Note
The number of mean vectors, covariance vectors and weights must be the same

Definition at line 89 of file mog_generic.cpp.

References check_array_uniformity(), D, do_checks, full, it_assert, K, paranoid, set_diag_covs_internal(), set_means_internal(), set_weights_internal(), setup_misc(), itpp::Array< T >::size(), and valid.

void itpp::MOG_generic::init ( Array< vec > &  means_in,
Array< mat > &  full_covs_in,
vec &  weights_in 
)

Initialise the model using user supplied parameters (full covariance version)

Parameters
means_inArray of mean vectors
full_covs_inArray of covariance matrices
weights_invector of weights
Note
The number of mean vectors, covariance matrices and weights must be the same

Definition at line 110 of file mog_generic.cpp.

References check_array_uniformity(), D, do_checks, full, it_assert, K, paranoid, set_full_covs_internal(), set_means_internal(), set_weights_internal(), setup_misc(), itpp::Array< T >::size(), and valid.

void itpp::MOG_generic::cleanup ( )
virtual

Release memory used by the model. The model will be empty.

Note
The likelihood functions are not useable until the model's parameters are re-initialised

Reimplemented in itpp::MOG_diag.

Definition at line 329 of file mog_generic.cpp.

References D, diag_covs, diag_covs_inv_etc, do_checks, full_covs, full_covs_inv, K, log_det_etc, log_weights, means, itpp::Array< T >::set_size(), valid, and weights.

Referenced by itpp::MOG_diag::cleanup(), and init().

void itpp::MOG_generic::set_means ( Array< vec > &  means_in)

Set the means of the model.

Note
The number of means must match the number of Gaussians in the model

Definition at line 351 of file mog_generic.cpp.

References set_means_internal(), and valid.

Referenced by init().

void itpp::MOG_generic::set_diag_covs ( Array< vec > &  diag_covs_in)

Set the diagonal covariance vectors of the model.

Note
The number of diagonal covariance vectors must match the number of Gaussians in the model

Definition at line 365 of file mog_generic.cpp.

References set_diag_covs_internal(), and valid.

void itpp::MOG_generic::set_full_covs ( Array< mat > &  full_covs_in)

Set the full covariance matrices of the model.

Note
The number of covariance matrices must match the number of Gaussians in the model

Definition at line 372 of file mog_generic.cpp.

References set_full_covs_internal(), and valid.

void itpp::MOG_generic::set_weights ( vec &  weights_in)

Set the weight vector of the model.

Note
The number of elements in the weight vector must match the number of Gaussians in the model

Definition at line 379 of file mog_generic.cpp.

References set_weights_internal(), and valid.

void itpp::MOG_generic::set_checks ( bool  do_checks_in)
inline

Enable/disable internal checks for likelihood functions.

Parameters
do_checks_inIf true, checks are enabled; if false, checks are disabled
Note
Disabling checks will provide a speedup in the likelihood functions. Disable them only when you're happy that everything is working correctly.

Definition at line 214 of file mog_generic.h.

void itpp::MOG_generic::set_paranoid ( bool  paranoid_in)
inline

Enable/disable paranoia about numerical stability.

Parameters
paranoid_inIf true, calculate likelihoods using a safer, but slower method.

Definition at line 219 of file mog_generic.h.

void itpp::MOG_generic::load ( const std::string &  name_in)
virtual

Initialise the model by loading the parameters from a model file.

Parameters
name_inThe model's filename

Reimplemented in itpp::MOG_diag.

Definition at line 407 of file mog_generic.cpp.

References itpp::it_file::close(), itpp::exist(), itpp::it_file::exists(), init(), it_assert, and valid.

Referenced by itpp::MOG_diag::load().

void itpp::MOG_generic::save ( const std::string &  name_in) const
virtual

Save the model's parameters to a model file.

Parameters
name_inThe model's filename

Definition at line 438 of file mog_generic.cpp.

References itpp::it_file::close(), diag_covs, full, full_covs, means, valid, and weights.

void itpp::MOG_generic::join ( const MOG_generic B_in)
virtual

Mathematically join the model with a user supplied model.

\param B_in user supplied model
\note The Arrays of mean vectors and covariance vectors/matrices from the two models
      are simply concatenated, while the weights of the resultant model are a function
      of the original weights and numbers of Gaussians from both models.
      Specifically,

$ w_{new} = [ \alpha \cdot w_{A} ~~~ \beta \cdot w_{B} ]^T $, where $ w_{new} $ is the new weight vector, $ w_{A} $ and $ w_{B} $ are the weight vectors from model A and B, while $ \alpha = K_A / (K_A + KB_in) $ and $ \beta = 1-\alpha $. In turn, $ K_A $ and $ KB_in $ are the numbers of Gaussians in model A and B, respectively.

See On transforming statistical models... for more information.

Definition at line 453 of file mog_generic.cpp.

References itpp::concat(), D, diag_covs, full, full_covs, get_D(), get_diag_covs(), get_full_covs(), get_K(), get_means(), get_weights(), init(), is_full(), is_valid(), it_assert, K, means, valid, and weights.

void itpp::MOG_generic::convert_to_diag ( )
virtual

Convert the model to use diagonal covariances.

Note
If the model is already diagonal, nothing is done. If the model has full covariance matrices, this results in irreversible information loss (in effect the off-diagonal covariance elements are now zero)

Definition at line 499 of file mog_generic.cpp.

References convert_to_diag_internal(), and valid.

Referenced by itpp::MOG_diag::load().

void itpp::MOG_generic::convert_to_full ( )
virtual

Convert the model to have full covariance matrices.

Note
If the model has full covariance matrices, nothing is done. If the model has diagonal covariances, the off-diagonal elements in the full covariance matrices are set to zero.

Reimplemented in itpp::MOG_diag.

Definition at line 518 of file mog_generic.cpp.

References convert_to_full_internal(), and valid.

Member Data Documentation

vec itpp::MOG_generic::log_det_etc
protected

Gaussian specific pre-calcualted constants.

Note
Vector of pre-calculated $ -\frac{D}{2}\log(2\pi) -\frac{1}{2}\log(|\Sigma|) $ for each Gaussian, where $ D $ is the dimensionality and $ |\Sigma| $ is the determinant for the Gaussian's covariance matrix $ \Sigma $.

Definition at line 317 of file mog_generic.h.

Referenced by cleanup(), log_lhood_single_gaus_internal(), itpp::MOG_diag::setup_covs(), and setup_covs().


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

Generated on Sat May 25 2013 16:32:34 for IT++ by Doxygen 1.8.2