SHOGUN  3.2.1
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
List of all members | Public Member Functions | Public Attributes | Protected Member Functions | Static Protected Member Functions | Protected Attributes
CKLDualInferenceMethod Class Reference

Detailed Description

The dual KL approximation inference method class.

This inference process is described in the reference paper Mohammad Emtiyaz Khan, Aleksandr Y. Aravkin, Michael P. Friedlander, Matthias Seeger Fast Dual Variational Inference for Non-Conjugate Latent Gaussian Models. ICML2013

The idea is to optimize the log marginal likelihood with equality constaints (primal problem) by solving the Lagrangian dual problem. The equality constaints are:

\[ h = \mu, \rho = \sigma^2 = diag(\Sigma) \]

, where h and \(\rho\) are auxiliary variables, \(\mu\) and \(\sigma^2\) are variational variables, and \(\Sigma\) is an approximated posterior covariance matrix. The equality constaints are variational mean constaint ( \(\mu\)) and variational variance constaint ( \(\sigma^2\)).

For detailed information, please refer to the paper.

Definition at line 65 of file KLDualInferenceMethod.h.

Inheritance diagram for CKLDualInferenceMethod:
Inheritance graph
[legend]

Public Member Functions

 CKLDualInferenceMethod ()
 CKLDualInferenceMethod (CKernel *kernel, CFeatures *features, CMeanFunction *mean, CLabels *labels, CLikelihoodModel *model)
virtual ~CKLDualInferenceMethod ()
virtual const char * get_name () const
virtual SGVector< float64_tget_alpha ()
virtual SGVector< float64_tget_diagonal_vector ()
void set_model (CLikelihoodModel *mod)
virtual EInferenceType get_inference_type () const
virtual float64_t get_negative_log_marginal_likelihood ()
virtual SGVector< float64_tget_posterior_mean ()
virtual SGMatrix< float64_tget_posterior_covariance ()
virtual bool supports_regression () const
virtual bool supports_binary () const
virtual void update ()
virtual void set_lbfgs_parameters (int m=100, int max_linesearch=1000, int linesearch=LBFGS_LINESEARCH_BACKTRACKING_STRONG_WOLFE, int max_iterations=1000, float64_t delta=0.0, int past=0, float64_t epsilon=1e-5, float64_t min_step=1e-20, float64_t max_step=1e+20, float64_t ftol=1e-4, float64_t wolfe=0.9, float64_t gtol=0.9, float64_t xtol=1e-16, float64_t orthantwise_c=0.0, int orthantwise_start=0, int orthantwise_end=1)
virtual SGMatrix< float64_tget_cholesky ()
virtual void set_noise_factor (float64_t noise_factor)
virtual void set_max_attempt (index_t max_attempt)
virtual void set_exp_factor (float64_t exp_factor)
virtual void set_min_coeff_kernel (float64_t min_coeff_kernel)
float64_t get_marginal_likelihood_estimate (int32_t num_importance_samples=1, float64_t ridge_size=1e-15)
virtual CMap< TParameter
*, SGVector< float64_t > > * 
get_negative_log_marginal_likelihood_derivatives (CMap< TParameter *, CSGObject * > *parameters)
virtual CMap< TParameter
*, SGVector< float64_t > > * 
get_gradient (CMap< TParameter *, CSGObject * > *parameters)
virtual SGVector< float64_tget_value ()
virtual CFeaturesget_features ()
virtual void set_features (CFeatures *feat)
virtual CKernelget_kernel ()
virtual void set_kernel (CKernel *kern)
virtual CMeanFunctionget_mean ()
virtual void set_mean (CMeanFunction *m)
virtual CLabelsget_labels ()
virtual void set_labels (CLabels *lab)
CLikelihoodModelget_model ()
virtual float64_t get_scale () const
virtual void set_scale (float64_t scale)
virtual bool supports_multiclass () const
virtual SGMatrix< float64_tget_multiclass_E ()
virtual CSGObjectshallow_copy () const
virtual CSGObjectdeep_copy () const
virtual bool is_generic (EPrimitiveType *generic) const
template<class T >
void set_generic ()
void unset_generic ()
virtual void print_serializable (const char *prefix="")
virtual bool save_serializable (CSerializableFile *file, const char *prefix="", int32_t param_version=Version::get_version_parameter())
virtual bool load_serializable (CSerializableFile *file, const char *prefix="", int32_t param_version=Version::get_version_parameter())
DynArray< TParameter * > * load_file_parameters (const SGParamInfo *param_info, int32_t file_version, CSerializableFile *file, const char *prefix="")
DynArray< TParameter * > * load_all_file_parameters (int32_t file_version, int32_t current_version, CSerializableFile *file, const char *prefix="")
void map_parameters (DynArray< TParameter * > *param_base, int32_t &base_version, DynArray< const SGParamInfo * > *target_param_infos)
void set_global_io (SGIO *io)
SGIOget_global_io ()
void set_global_parallel (Parallel *parallel)
Parallelget_global_parallel ()
void set_global_version (Version *version)
Versionget_global_version ()
SGStringList< char > get_modelsel_names ()
void print_modsel_params ()
char * get_modsel_param_descr (const char *param_name)
index_t get_modsel_param_index (const char *param_name)
void build_gradient_parameter_dictionary (CMap< TParameter *, CSGObject * > *dict)
virtual void update_parameter_hash ()
virtual bool parameter_hash_changed ()
virtual bool equals (CSGObject *other, float64_t accuracy=0.0, bool tolerant=false)
virtual CSGObjectclone ()

Public Attributes

SGIOio
Parallelparallel
Versionversion
Parameterm_parameters
Parameterm_model_selection_parameters
Parameterm_gradient_parameters
ParameterMapm_parameter_map
uint32_t m_hash

Protected Member Functions

virtual void get_gradient_of_nlml_wrt_parameters (SGVector< float64_t > gradient)
virtual
CDualVariationalGaussianLikelihood
get_dual_variational_likelihood () const
virtual void check_dual_inference (CLikelihoodModel *mod) const
virtual void update_approx_cov ()
virtual void update_alpha ()
virtual void update_chol ()
virtual void update_deriv ()
virtual float64_t get_negative_log_marginal_likelihood_helper ()
virtual bool lbfgs_precompute ()
virtual float64_t get_derivative_related_cov (Eigen::MatrixXd eigen_dK)
virtual float64_t lbfgs_optimization ()
virtual float64_t get_dual_objective_wrt_parameters ()
virtual void get_gradient_of_dual_objective_wrt_parameters (SGVector< float64_t > gradient)
virtual void update_init ()
virtual Eigen::LDLT
< Eigen::MatrixXd, 0x1 > 
update_init_helper ()
virtual
CVariationalGaussianLikelihood
get_variational_likelihood () const
virtual void check_variational_likelihood (CLikelihoodModel *mod) const
virtual SGVector< float64_tget_derivative_wrt_inference_method (const TParameter *param)
virtual SGVector< float64_tget_derivative_wrt_likelihood_model (const TParameter *param)
virtual SGVector< float64_tget_derivative_wrt_kernel (const TParameter *param)
virtual SGVector< float64_tget_derivative_wrt_mean (const TParameter *param)
virtual float64_t get_nlml_wrt_parameters ()
virtual void check_members () const
virtual void update_train_kernel ()
virtual TParametermigrate (DynArray< TParameter * > *param_base, const SGParamInfo *target)
virtual void one_to_one_migration_prepare (DynArray< TParameter * > *param_base, const SGParamInfo *target, TParameter *&replacement, TParameter *&to_migrate, char *old_name=NULL)
virtual void load_serializable_pre () throw (ShogunException)
virtual void load_serializable_post () throw (ShogunException)
virtual void save_serializable_pre () throw (ShogunException)
virtual void save_serializable_post () throw (ShogunException)

Static Protected Member Functions

static void * get_derivative_helper (void *p)

Protected Attributes

float64_t m_min_coeff_kernel
float64_t m_noise_factor
float64_t m_exp_factor
index_t m_max_attempt
SGVector< float64_tm_mu
SGMatrix< float64_tm_Sigma
SGVector< float64_tm_s2
int m_m
int m_max_linesearch
int m_linesearch
int m_max_iterations
float64_t m_delta
int m_past
float64_t m_epsilon
float64_t m_min_step
float64_t m_max_step
float64_t m_ftol
float64_t m_wolfe
float64_t m_gtol
float64_t m_xtol
float64_t m_orthantwise_c
int m_orthantwise_start
int m_orthantwise_end
CKernelm_kernel
CMeanFunctionm_mean
CLikelihoodModelm_model
CFeaturesm_features
CLabelsm_labels
SGVector< float64_tm_alpha
SGMatrix< float64_tm_L
float64_t m_scale
SGMatrix< float64_tm_ktrtr
SGMatrix< float64_tm_E

Constructor & Destructor Documentation

default constructor

Definition at line 55 of file KLDualInferenceMethod.cpp.

CKLDualInferenceMethod ( CKernel kernel,
CFeatures features,
CMeanFunction mean,
CLabels labels,
CLikelihoodModel model 
)

constructor

Parameters
kernelcovariance function
featuresfeatures to use in inference
meanmean function
labelslabels of the features
modelLikelihood model to use

Definition at line 60 of file KLDualInferenceMethod.cpp.

~CKLDualInferenceMethod ( )
virtual

Definition at line 76 of file KLDualInferenceMethod.cpp.

Member Function Documentation

void build_gradient_parameter_dictionary ( CMap< TParameter *, CSGObject * > *  dict)
inherited

Builds a dictionary of all parameters in SGObject as well of those of SGObjects that are parameters of this object. Dictionary maps parameters to the objects that own them.

Parameters
dictdictionary of parameters to be built.

Definition at line 1243 of file SGObject.cpp.

void check_dual_inference ( CLikelihoodModel mod) const
protectedvirtual

check the provided likelihood model supports dual variational inference or not

Parameters
modthe provided likelihood model
Returns
whether the provided likelihood model supports dual variational inference or not

Definition at line 80 of file KLDualInferenceMethod.cpp.

void check_members ( ) const
protectedvirtualinherited

check if members of object are valid for inference

Reimplemented in CFITCInferenceMethod, and CExactInferenceMethod.

Definition at line 275 of file InferenceMethod.cpp.

void check_variational_likelihood ( CLikelihoodModel mod) const
protectedvirtualinherited

check the provided likelihood model supports variational inference

Parameters
modthe provided likelihood model
Returns
whether the provided likelihood model supports variational inference or not

Definition at line 57 of file KLInferenceMethod.cpp.

CSGObject * clone ( )
virtualinherited

Creates a clone of the current object. This is done via recursively traversing all parameters, which corresponds to a deep copy. Calling equals on the cloned object always returns true although none of the memory of both objects overlaps.

Returns
an identical copy of the given object, which is disjoint in memory. NULL if the clone fails. Note that the returned object is SG_REF'ed

Definition at line 1360 of file SGObject.cpp.

CSGObject * deep_copy ( ) const
virtualinherited

A deep copy. All the instance variables will also be copied.

Definition at line 200 of file SGObject.cpp.

bool equals ( CSGObject other,
float64_t  accuracy = 0.0,
bool  tolerant = false 
)
virtualinherited

Recursively compares the current SGObject to another one. Compares all registered numerical parameters, recursion upon complex (SGObject) parameters. Does not compare pointers!

May be overwritten but please do with care! Should not be necessary in most cases.

Parameters
otherobject to compare with
accuracyaccuracy to use for comparison (optional)
tolerantallows linient check on float equality (within accuracy)
Returns
true if all parameters were equal, false if not

Definition at line 1264 of file SGObject.cpp.

SGVector< float64_t > get_alpha ( )
virtual

get alpha vector

Returns
vector to compute posterior mean of Gaussian Process:

\[ \mu = K\alpha+mean_f \]

where \(\mu\) is the mean and \(K\) is the prior covariance matrix.

Definition at line 67 of file KLDualInferenceMethod.cpp.

SGMatrix< float64_t > get_cholesky ( )
virtualinherited

get Cholesky decomposition matrix

Returns
Cholesky decomposition of matrix:

\[ L = cholesky(sW*K*sW+I) \]

where \(K\) is the prior covariance matrix, \(sW\) is the vector returned by get_diagonal_vector(), and \(I\) is the identity matrix.

Note that in some sub class L is not the Cholesky decomposition In this case, L will still be used to compute required matrix for prediction see CGaussianProcessMachine::get_posterior_variances()

Definition at line 461 of file KLInferenceMethod.cpp.

void * get_derivative_helper ( void *  p)
staticprotectedinherited

pthread helper method to compute negative log marginal likelihood derivatives wrt hyperparameter

Definition at line 221 of file InferenceMethod.cpp.

float64_t get_derivative_related_cov ( Eigen::MatrixXd  eigen_dK)
protectedvirtual

compute matrices which are required to compute negative log marginal likelihood derivatives wrt hyperparameter in cov function Note that get_derivative_wrt_inference_method(const TParameter* param) and get_derivative_wrt_kernel(const TParameter* param) will call this function

Parameters
thegradient related to cov
Returns
the gradient wrt hyperparameter related to cov

Implements CKLInferenceMethod.

Definition at line 245 of file KLDualInferenceMethod.cpp.

SGVector< float64_t > get_derivative_wrt_inference_method ( const TParameter param)
protectedvirtualinherited

returns derivative of negative log marginal likelihood wrt parameter of CInferenceMethod class

Parameters
paramparameter of CInferenceMethod class
Returns
derivative of negative log marginal likelihood

Implements CInferenceMethod.

Definition at line 410 of file KLInferenceMethod.cpp.

SGVector< float64_t > get_derivative_wrt_kernel ( const TParameter param)
protectedvirtualinherited

returns derivative of negative log marginal likelihood wrt kernel's parameter

Parameters
paramparameter of given kernel
Returns
derivative of negative log marginal likelihood

Implements CInferenceMethod.

Definition at line 427 of file KLInferenceMethod.cpp.

SGVector< float64_t > get_derivative_wrt_likelihood_model ( const TParameter param)
protectedvirtualinherited

returns derivative of negative log marginal likelihood wrt parameter of likelihood model

Parameters
paramparameter of given likelihood model
Returns
derivative of negative log marginal likelihood

Implements CInferenceMethod.

Definition at line 326 of file KLInferenceMethod.cpp.

SGVector< float64_t > get_derivative_wrt_mean ( const TParameter param)
protectedvirtualinherited

returns derivative of negative log marginal likelihood wrt mean function's parameter

Parameters
paramparameter of given mean function
Returns
derivative of negative log marginal likelihood

Implements CInferenceMethod.

Definition at line 342 of file KLInferenceMethod.cpp.

SGVector< float64_t > get_diagonal_vector ( )
virtual

get diagonal vector

Returns
diagonal of matrix used to calculate posterior covariance matrix:

\[ Cov = (K^{-1}+sW^{2})^{-1} \]

where \(Cov\) is the posterior covariance matrix, \(K\) is the prior covariance matrix, and \(sW\) is the diagonal vector.

Definition at line 426 of file KLDualInferenceMethod.cpp.

float64_t get_dual_objective_wrt_parameters ( )
protectedvirtual

compute the objective value for LBFGS optimizer

The mathematical equation (equation 24 in the paper) is defined as below

\[ min_{\lambda \in S}{0.5*[(\lambda-y)^TK(\lambda-y)-log(det(A_{\lambda}))]-mean_{f}^T(\lambda-y)+\sum_{i=1}^{n}{Fenchel_i{(\lambda)}}} \]

where S is the feasible set defined for \(\lambda\), K comes from covariance function, \(mean_f\) comes from mean function, \(\lambda\) is the dual parameter, y are data labels, n is the number point, \(A_{\lambda}=K^{-1}+diag(\lambda)\), and \(Fenchel_i{(\lambda)}=Fenchel_i{(\alpha,\lambda)}\) since \(\alpha\) is implicitly defined by \(\lambda\)

Note that S and \(Fenchel_i{(\lambda)}\) are specified by the data modeling distribution, which are implemented in dual variational likelihood class.

Definition at line 160 of file KLDualInferenceMethod.cpp.

CDualVariationalGaussianLikelihood * get_dual_variational_likelihood ( ) const
protectedvirtual

this method is used to dynamic-cast the likelihood model, m_model, to dual variational likelihood model.

Definition at line 93 of file KLDualInferenceMethod.cpp.

virtual CFeatures* get_features ( )
virtualinherited

get features

Returns
features

Definition at line 236 of file InferenceMethod.h.

SGIO * get_global_io ( )
inherited

get the io object

Returns
io object

Definition at line 237 of file SGObject.cpp.

Parallel * get_global_parallel ( )
inherited

get the parallel object

Returns
parallel object

Definition at line 278 of file SGObject.cpp.

Version * get_global_version ( )
inherited

get the version object

Returns
version object

Definition at line 291 of file SGObject.cpp.

virtual CMap<TParameter*, SGVector<float64_t> >* get_gradient ( CMap< TParameter *, CSGObject * > *  parameters)
virtualinherited

get the gradient

Parameters
parametersparameter's dictionary
Returns
map of gradient. Keys are names of parameters, values are values of derivative with respect to that parameter.

Implements CDifferentiableFunction.

Definition at line 215 of file InferenceMethod.h.

void get_gradient_of_dual_objective_wrt_parameters ( SGVector< float64_t gradient)
protectedvirtual

compute the gradient of the objective function for LBFGS optimizer The mathematical equation (equation 25 in the paper) is defined as below

\[ 0.5*[2*K(\lambda-y)-diag(A_{\lambda}^{-1})]-mean_{f}+\sum_{i=1}^{n}{\nabla Fenchel_i{(\lambda)}} \]

where \(A_{\lambda}=K^{-1}+diag(\lambda)\), K comes from covariance function, \(mean_f\) comes from mean function, \(\lambda\) is the dual parameter, y are data labels, n is the number point, and \(\nabla Fenchel_i{(\lambda)}\) is the gradient of \(Fenchel_i{(\lambda)}\) wrt to \(\lambda\)

*Note that \(\nabla Fenchel_i{(\lambda)}\) are specified by the data modeling distribution, *which are implemented in dual variational likelihood class.

Definition at line 181 of file KLDualInferenceMethod.cpp.

virtual void get_gradient_of_nlml_wrt_parameters ( SGVector< float64_t gradient)
protectedvirtual

compute the gradient wrt variational parameters given the current variational parameters (mu and s2)

Returns
gradient of negative log marginal likelihood

Implements CKLInferenceMethod.

Definition at line 127 of file KLDualInferenceMethod.h.

virtual EInferenceType get_inference_type ( ) const
virtualinherited

return what type of inference we are

Reimplemented from CInferenceMethod.

Definition at line 99 of file KLInferenceMethod.h.

virtual CKernel* get_kernel ( )
virtualinherited

get kernel

Returns
kernel

Definition at line 253 of file InferenceMethod.h.

virtual CLabels* get_labels ( )
virtualinherited

get labels

Returns
labels

Definition at line 287 of file InferenceMethod.h.

float64_t get_marginal_likelihood_estimate ( int32_t  num_importance_samples = 1,
float64_t  ridge_size = 1e-15 
)
inherited

Computes an unbiased estimate of the marginal-likelihood (in log-domain),

\[ p(y|X,\theta), \]

where \(y\) are the labels, \(X\) are the features (omitted from in the following expressions), and \(\theta\) represent hyperparameters.

This is done via a Gaussian approximation to the posterior \(q(f|y, \theta)\approx p(f|y, \theta)\), which is computed by the underlying CInferenceMethod instance (if implemented, otherwise error), and then using an importance sample estimator

\[ p(y|\theta)=\int p(y|f)p(f|\theta)df =\int p(y|f)\frac{p(f|\theta)}{q(f|y, \theta)}q(f|y, \theta)df \approx\frac{1}{n}\sum_{i=1}^n p(y|f^{(i)})\frac{p(f^{(i)}|\theta)} {q(f^{(i)}|y, \theta)}, \]

where \( f^{(i)} \) are samples from the posterior approximation \( q(f|y, \theta) \). The resulting estimator has a low variance if \( q(f|y, \theta) \) is a good approximation. It has large variance otherwise (while still being consistent). Storing all number of log-domain ensures numerical stability.

Parameters
num_importance_samplesthe number of importance samples \(n\) from \( q(f|y, \theta) \).
ridge_sizescalar that is added to the diagonal of the involved Gaussian distribution's covariance of GP prior and posterior approximation to stabilise things. Increase if covariance matrix is not numerically positive semi-definite.
Returns
unbiased estimate of the marginal likelihood function \( p(y|\theta),\) in log-domain.

Definition at line 91 of file InferenceMethod.cpp.

virtual CMeanFunction* get_mean ( )
virtualinherited

get mean

Returns
mean

Definition at line 270 of file InferenceMethod.h.

CLikelihoodModel* get_model ( )
inherited

get likelihood model

Returns
likelihood

Definition at line 304 of file InferenceMethod.h.

SGStringList< char > get_modelsel_names ( )
inherited
Returns
vector of names of all parameters which are registered for model selection

Definition at line 1135 of file SGObject.cpp.

char * get_modsel_param_descr ( const char *  param_name)
inherited

Returns description of a given parameter string, if it exists. SG_ERROR otherwise

Parameters
param_namename of the parameter
Returns
description of the parameter

Definition at line 1159 of file SGObject.cpp.

index_t get_modsel_param_index ( const char *  param_name)
inherited

Returns index of model selection parameter with provided index

Parameters
param_namename of model selection parameter
Returns
index of model selection parameter with provided name, -1 if there is no such

Definition at line 1172 of file SGObject.cpp.

SGMatrix< float64_t > get_multiclass_E ( )
virtualinherited

get the E matrix used for multi classification

Returns
the matrix for multi classification

Definition at line 40 of file InferenceMethod.cpp.

virtual const char* get_name ( ) const
virtual

returns the name of the inference method

Returns
name KLDualInferenceMethod

Reimplemented from CKLInferenceMethod.

Definition at line 88 of file KLDualInferenceMethod.h.

float64_t get_negative_log_marginal_likelihood ( )
virtualinherited

get negative log marginal likelihood

Returns
the negative log of the marginal likelihood function:

\[ -log(p(y|X, \theta)) \]

where \(y\) are the labels, \(X\) are the features, and \(\theta\) represent hyperparameters.

Implements CInferenceMethod.

Definition at line 318 of file KLInferenceMethod.cpp.

CMap< TParameter *, SGVector< float64_t > > * get_negative_log_marginal_likelihood_derivatives ( CMap< TParameter *, CSGObject * > *  parameters)
virtualinherited

get log marginal likelihood gradient

Returns
vector of the marginal likelihood function gradient with respect to hyperparameters (under the current approximation to the posterior \(q(f|y)\approx p(f|y)\):

\[ -\frac{\partial log(p(y|X, \theta))}{\partial \theta} \]

where \(y\) are the labels, \(X\) are the features, and \(\theta\) represent hyperparameters.

Definition at line 150 of file InferenceMethod.cpp.

float64_t get_negative_log_marginal_likelihood_helper ( )
protectedvirtual

the helper function to compute the negative log marginal likelihood

Returns
negative log marginal likelihood

Implements CKLInferenceMethod.

Definition at line 236 of file KLDualInferenceMethod.cpp.

float64_t get_nlml_wrt_parameters ( )
protectedvirtualinherited

compute the negative log marginal likelihood given the current variational parameters (mu and s2)

Returns
negative log marginal likelihood

Definition at line 275 of file KLInferenceMethod.cpp.

SGMatrix< float64_t > get_posterior_covariance ( )
virtualinherited

returns covariance matrix \(\Sigma=(K^{-1}+W)^{-1}\) of the Gaussian distribution \(\mathcal{N}(\mu,\Sigma)\), which is an approximation to the posterior:

\[ p(f|y) \approx q(f|y) = \mathcal{N}(f|\mu,\Sigma) \]

Covariance matrix is evaluated using matrix inversion lemma:

\[ (K^{-1}+W)^{-1} = K - KW^{\frac{1}{2}}B^{-1}W^{\frac{1}{2}}K \]

where \(B=(W^{frac{1}{2}}*K*W^{frac{1}{2}}+I)\).

Returns
covariance matrix

Implements CInferenceMethod.

Definition at line 239 of file KLInferenceMethod.cpp.

SGVector< float64_t > get_posterior_mean ( )
virtualinherited

returns mean vector \(\mu\) of the Gaussian distribution \(\mathcal{N}(\mu,\Sigma)\), which is an approximation to the posterior:

\[ p(f|y) \approx q(f|y) = \mathcal{N}(f|\mu,\Sigma) \]

Returns
mean vector

Implements CInferenceMethod.

Definition at line 231 of file KLInferenceMethod.cpp.

virtual float64_t get_scale ( ) const
virtualinherited

get kernel scale

Returns
kernel scale

Definition at line 321 of file InferenceMethod.h.

virtual SGVector<float64_t> get_value ( )
virtualinherited

get the function value

Returns
vector that represents the function value

Implements CDifferentiableFunction.

Definition at line 225 of file InferenceMethod.h.

CVariationalGaussianLikelihood * get_variational_likelihood ( ) const
protectedvirtualinherited

this method is used to dynamic-cast the likelihood model, m_model, to variational likelihood model.

Definition at line 268 of file KLInferenceMethod.cpp.

bool is_generic ( EPrimitiveType *  generic) const
virtualinherited

If the SGSerializable is a class template then TRUE will be returned and GENERIC is set to the type of the generic.

Parameters
genericset to the type of the generic if returning TRUE
Returns
TRUE if a class template.

Definition at line 297 of file SGObject.cpp.

float64_t lbfgs_optimization ( )
protectedvirtual

Using L-BFGS to estimate posterior parameters

Reimplemented from CKLInferenceMethod.

Definition at line 396 of file KLDualInferenceMethod.cpp.

bool lbfgs_precompute ( )
protectedvirtual

pre-compute the information for lbfgs optimization. This function needs to be called before calling get_negative_log_marginal_likelihood_wrt_parameters() and/or get_gradient_of_nlml_wrt_parameters(SGVector<float64_t> gradient)

Returns
true if precomputed parameters are valid

Implements CKLInferenceMethod.

Definition at line 120 of file KLDualInferenceMethod.cpp.

DynArray< TParameter * > * load_all_file_parameters ( int32_t  file_version,
int32_t  current_version,
CSerializableFile file,
const char *  prefix = "" 
)
inherited

maps all parameters of this instance to the provided file version and loads all parameter data from the file into an array, which is sorted (basically calls load_file_parameter(...) for all parameters and puts all results into a sorted array)

Parameters
file_versionparameter version of the file
current_versionversion from which mapping begins (you want to use Version::get_version_parameter() for this in most cases)
filefile to load from
prefixprefix for members
Returns
(sorted) array of created TParameter instances with file data

Definition at line 704 of file SGObject.cpp.

DynArray< TParameter * > * load_file_parameters ( const SGParamInfo param_info,
int32_t  file_version,
CSerializableFile file,
const char *  prefix = "" 
)
inherited

loads some specified parameters from a file with a specified version The provided parameter info has a version which is recursively mapped until the file parameter version is reached. Note that there may be possibly multiple parameters in the mapping, therefore, a set of TParameter instances is returned

Parameters
param_infoinformation of parameter
file_versionparameter version of the file, must be <= provided parameter version
filefile to load from
prefixprefix for members
Returns
new array with TParameter instances with the attached data

Definition at line 545 of file SGObject.cpp.

bool load_serializable ( CSerializableFile file,
const char *  prefix = "",
int32_t  param_version = Version::get_version_parameter() 
)
virtualinherited

Load this object from file. If it will fail (returning FALSE) then this object will contain inconsistent data and should not be used!

Parameters
filewhere to load from
prefixprefix for members
param_version(optional) a parameter version different to (this is mainly for testing, better do not use)
Returns
TRUE if done, otherwise FALSE

Definition at line 374 of file SGObject.cpp.

void load_serializable_post ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to post-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::LOAD_SERIALIZABLE_POST is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CKernel, CWeightedDegreePositionStringKernel, CList, CAlphabet, CLinearHMM, CGaussianKernel, CInverseMultiQuadricKernel, CCircularKernel, and CExponentialKernel.

Definition at line 1062 of file SGObject.cpp.

void load_serializable_pre ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to pre-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::LOAD_SERIALIZABLE_PRE is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CDynamicArray< T >, CDynamicArray< float64_t >, CDynamicArray< float32_t >, CDynamicArray< int32_t >, CDynamicArray< char >, CDynamicArray< bool >, and CDynamicObjectArray.

Definition at line 1057 of file SGObject.cpp.

void map_parameters ( DynArray< TParameter * > *  param_base,
int32_t &  base_version,
DynArray< const SGParamInfo * > *  target_param_infos 
)
inherited

Takes a set of TParameter instances (base) with a certain version and a set of target parameter infos and recursively maps the base level wise to the current version using CSGObject::migrate(...). The base is replaced. After this call, the base version containing parameters should be of same version/type as the initial target parameter infos. Note for this to work, the migrate methods and all the internal parameter mappings have to match

Parameters
param_baseset of TParameter instances that are mapped to the provided target parameter infos
base_versionversion of the parameter base
target_param_infosset of SGParamInfo instances that specify the target parameter base

Definition at line 742 of file SGObject.cpp.

TParameter * migrate ( DynArray< TParameter * > *  param_base,
const SGParamInfo target 
)
protectedvirtualinherited

creates a new TParameter instance, which contains migrated data from the version that is provided. The provided parameter data base is used for migration, this base is a collection of all parameter data of the previous version. Migration is done FROM the data in param_base TO the provided param info Migration is always one version step. Method has to be implemented in subclasses, if no match is found, base method has to be called.

If there is an element in the param_base which equals the target, a copy of the element is returned. This represents the case when nothing has changed and therefore, the migrate method is not overloaded in a subclass

Parameters
param_baseset of TParameter instances to use for migration
targetparameter info for the resulting TParameter
Returns
a new TParameter instance with migrated data from the base of the type which is specified by the target parameter

Definition at line 949 of file SGObject.cpp.

void one_to_one_migration_prepare ( DynArray< TParameter * > *  param_base,
const SGParamInfo target,
TParameter *&  replacement,
TParameter *&  to_migrate,
char *  old_name = NULL 
)
protectedvirtualinherited

This method prepares everything for a one-to-one parameter migration. One to one here means that only ONE element of the parameter base is needed for the migration (the one with the same name as the target). Data is allocated for the target (in the type as provided in the target SGParamInfo), and a corresponding new TParameter instance is written to replacement. The to_migrate pointer points to the single needed TParameter instance needed for migration. If a name change happened, the old name may be specified by old_name. In addition, the m_delete_data flag of to_migrate is set to true. So if you want to migrate data, the only thing to do after this call is converting the data in the m_parameter fields. If unsure how to use - have a look into an example for this. (base_migration_type_conversion.cpp for example)

Parameters
param_baseset of TParameter instances to use for migration
targetparameter info for the resulting TParameter
replacement(used as output) here the TParameter instance which is returned by migration is created into
to_migratethe only source that is used for migration
old_namewith this parameter, a name change may be specified

Definition at line 889 of file SGObject.cpp.

bool parameter_hash_changed ( )
virtualinherited
Returns
whether parameter combination has changed since last update

Definition at line 263 of file SGObject.cpp.

void print_modsel_params ( )
inherited

prints all parameter registered for model selection and their type

Definition at line 1111 of file SGObject.cpp.

void print_serializable ( const char *  prefix = "")
virtualinherited

prints registered parameters out

Parameters
prefixprefix for members

Definition at line 309 of file SGObject.cpp.

bool save_serializable ( CSerializableFile file,
const char *  prefix = "",
int32_t  param_version = Version::get_version_parameter() 
)
virtualinherited

Save this object to file.

Parameters
filewhere to save the object; will be closed during returning if PREFIX is an empty string.
prefixprefix for members
param_version(optional) a parameter version different to (this is mainly for testing, better do not use)
Returns
TRUE if done, otherwise FALSE

Definition at line 315 of file SGObject.cpp.

void save_serializable_post ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to post-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::SAVE_SERIALIZABLE_POST is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CKernel.

Definition at line 1072 of file SGObject.cpp.

void save_serializable_pre ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to pre-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::SAVE_SERIALIZABLE_PRE is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CKernel, CDynamicArray< T >, CDynamicArray< float64_t >, CDynamicArray< float32_t >, CDynamicArray< int32_t >, CDynamicArray< char >, CDynamicArray< bool >, and CDynamicObjectArray.

Definition at line 1067 of file SGObject.cpp.

void set_exp_factor ( float64_t  exp_factor)
virtualinherited

set exp factor to exponentially increase noise factor

Parameters
exp_factorshould be greater than 1.0 default value is 2

Definition at line 189 of file KLInferenceMethod.cpp.

virtual void set_features ( CFeatures feat)
virtualinherited

set features

Parameters
featfeatures to set

Definition at line 242 of file InferenceMethod.h.

void set_generic< complex128_t > ( )
inherited

set generic type to T

Definition at line 42 of file SGObject.cpp.

void set_global_io ( SGIO io)
inherited

set the io object

Parameters
ioio object to use

Definition at line 230 of file SGObject.cpp.

void set_global_parallel ( Parallel parallel)
inherited

set the parallel object

Parameters
parallelparallel object to use

Definition at line 243 of file SGObject.cpp.

void set_global_version ( Version version)
inherited

set the version object

Parameters
versionversion object to use

Definition at line 284 of file SGObject.cpp.

virtual void set_kernel ( CKernel kern)
virtualinherited

set kernel

Parameters
kernkernel to set

Definition at line 259 of file InferenceMethod.h.

virtual void set_labels ( CLabels lab)
virtualinherited

set labels

Parameters
lablabel to set

Definition at line 293 of file InferenceMethod.h.

void set_lbfgs_parameters ( int  m = 100,
int  max_linesearch = 1000,
int  linesearch = LBFGS_LINESEARCH_BACKTRACKING_STRONG_WOLFE,
int  max_iterations = 1000,
float64_t  delta = 0.0,
int  past = 0,
float64_t  epsilon = 1e-5,
float64_t  min_step = 1e-20,
float64_t  max_step = 1e+20,
float64_t  ftol = 1e-4,
float64_t  wolfe = 0.9,
float64_t  gtol = 0.9,
float64_t  xtol = 1e-16,
float64_t  orthantwise_c = 0.0,
int  orthantwise_start = 0,
int  orthantwise_end = 1 
)
virtualinherited

Definition at line 282 of file KLInferenceMethod.cpp.

void set_max_attempt ( index_t  max_attempt)
virtualinherited

set max attempt to ensure Kernel matrix to be positive definite

Parameters
max_attemptshould be non-negative. 0 means infinity attempts default value is 0

Definition at line 183 of file KLInferenceMethod.cpp.

virtual void set_mean ( CMeanFunction m)
virtualinherited

set mean

Parameters
mmean function to set

Definition at line 276 of file InferenceMethod.h.

void set_min_coeff_kernel ( float64_t  min_coeff_kernel)
virtualinherited

set minimum coeefficient of kernel matrix used in LDLT factorization

Parameters
min_coeff_kernelshould be non-negative default value is 1e-5

Definition at line 177 of file KLInferenceMethod.cpp.

void set_model ( CLikelihoodModel mod)
virtual

set variational likelihood model

Parameters
modmodel to set

Reimplemented from CKLInferenceMethod.

Definition at line 87 of file KLDualInferenceMethod.cpp.

void set_noise_factor ( float64_t  noise_factor)
virtualinherited

set noise factor to ensure Kernel matrix to be positive definite by adding non-negative noise to diagonal elements of Kernel matrix

Parameters
noise_factorshould be non-negative default value is 1e-10

Definition at line 171 of file KLInferenceMethod.cpp.

virtual void set_scale ( float64_t  scale)
virtualinherited

set kernel scale

Parameters
scalescale to be set

Definition at line 327 of file InferenceMethod.h.

CSGObject * shallow_copy ( ) const
virtualinherited

A shallow copy. All the SGObject instance variables will be simply assigned and SG_REF-ed.

Reimplemented in CGaussianKernel.

Definition at line 194 of file SGObject.cpp.

virtual bool supports_binary ( ) const
virtualinherited
Returns
whether combination of KL approximation inference method and given likelihood function supports binary classification

Reimplemented from CInferenceMethod.

Definition at line 167 of file KLInferenceMethod.h.

virtual bool supports_multiclass ( ) const
virtualinherited

whether combination of inference method and given likelihood function supports multiclass classification

Returns
false

Definition at line 348 of file InferenceMethod.h.

virtual bool supports_regression ( ) const
virtualinherited
Returns
whether combination of KL approximation inference method and given likelihood function supports regression

Reimplemented from CInferenceMethod.

Definition at line 157 of file KLInferenceMethod.h.

void unset_generic ( )
inherited

unset generic type

this has to be called in classes specializing a template class

Definition at line 304 of file SGObject.cpp.

void update ( )
virtualinherited

update data all matrices

Reimplemented from CInferenceMethod.

Definition at line 156 of file KLInferenceMethod.cpp.

void update_alpha ( )
protectedvirtual

update alpha matrix

Implements CInferenceMethod.

Definition at line 276 of file KLDualInferenceMethod.cpp.

void update_approx_cov ( )
protectedvirtual

update covariance matrix of the approximation to the posterior

Implements CKLInferenceMethod.

Definition at line 448 of file KLDualInferenceMethod.cpp.

void update_chol ( )
protectedvirtual

update cholesky matrix

Implements CInferenceMethod.

Definition at line 441 of file KLDualInferenceMethod.cpp.

void update_deriv ( )
protectedvirtual

update matrices which are required to compute negative log marginal likelihood derivatives wrt hyperparameter

Implements CInferenceMethod.

Definition at line 434 of file KLDualInferenceMethod.cpp.

void update_init ( )
protectedvirtualinherited

correct the kernel matrix and factorizated the corrected Kernel matrix for update

Reimplemented in CKLLowerTriangularInferenceMethod.

Definition at line 195 of file KLInferenceMethod.cpp.

Eigen::LDLT< Eigen::MatrixXd > update_init_helper ( )
protectedvirtualinherited

a helper function used to correct the kernel matrix using LDLT factorization

Returns
the LDLT factorization of the corrected kernel matrix

Definition at line 200 of file KLInferenceMethod.cpp.

void update_parameter_hash ( )
virtualinherited

Updates the hash of current parameter combination

Definition at line 250 of file SGObject.cpp.

void update_train_kernel ( )
protectedvirtualinherited

update train kernel matrix

Reimplemented in CFITCInferenceMethod.

Definition at line 291 of file InferenceMethod.cpp.

Member Data Documentation

SGIO* io
inherited

io

Definition at line 496 of file SGObject.h.

SGVector<float64_t> m_alpha
protectedinherited

alpha vector used in process mean calculation

Definition at line 443 of file InferenceMethod.h.

float64_t m_delta
protectedinherited

Definition at line 437 of file KLInferenceMethod.h.

SGMatrix<float64_t> m_E
protectedinherited

the matrix used for multi classification

Definition at line 455 of file InferenceMethod.h.

float64_t m_epsilon
protectedinherited

Definition at line 443 of file KLInferenceMethod.h.

float64_t m_exp_factor
protectedinherited

The factor used to exponentially increase noise_factor

Definition at line 294 of file KLInferenceMethod.h.

CFeatures* m_features
protectedinherited

features to use

Definition at line 437 of file InferenceMethod.h.

float64_t m_ftol
protectedinherited

Definition at line 452 of file KLInferenceMethod.h.

Parameter* m_gradient_parameters
inherited

parameters wrt which we can compute gradients

Definition at line 511 of file SGObject.h.

float64_t m_gtol
protectedinherited

Definition at line 458 of file KLInferenceMethod.h.

uint32_t m_hash
inherited

Hash of parameter values

Definition at line 517 of file SGObject.h.

CKernel* m_kernel
protectedinherited

covariance function

Definition at line 428 of file InferenceMethod.h.

SGMatrix<float64_t> m_ktrtr
protectedinherited

kernel matrix from features (non-scalled by inference scalling)

Definition at line 452 of file InferenceMethod.h.

SGMatrix<float64_t> m_L
protectedinherited

upper triangular factor of Cholesky decomposition

Definition at line 446 of file InferenceMethod.h.

CLabels* m_labels
protectedinherited

labels of features

Definition at line 440 of file InferenceMethod.h.

int m_linesearch
protectedinherited

Definition at line 431 of file KLInferenceMethod.h.

int m_m
protectedinherited

Definition at line 425 of file KLInferenceMethod.h.

index_t m_max_attempt
protectedinherited

Max number of attempt to correct kernel matrix to be positive definite

Definition at line 297 of file KLInferenceMethod.h.

int m_max_iterations
protectedinherited

Definition at line 434 of file KLInferenceMethod.h.

int m_max_linesearch
protectedinherited

Definition at line 428 of file KLInferenceMethod.h.

float64_t m_max_step
protectedinherited

Definition at line 449 of file KLInferenceMethod.h.

CMeanFunction* m_mean
protectedinherited

mean function

Definition at line 431 of file InferenceMethod.h.

float64_t m_min_coeff_kernel
protectedinherited

The minimum coeefficient of kernel matrix in LDLT factorization used to check whether the kernel matrix is positive definite or not

Definition at line 288 of file KLInferenceMethod.h.

float64_t m_min_step
protectedinherited

Definition at line 446 of file KLInferenceMethod.h.

CLikelihoodModel* m_model
protectedinherited

likelihood function to use

Definition at line 434 of file InferenceMethod.h.

Parameter* m_model_selection_parameters
inherited

model selection parameters

Definition at line 508 of file SGObject.h.

SGVector<float64_t> m_mu
protectedinherited

mean vector of the approximation to the posterior Note that m_mu is also a variational parameter

Definition at line 414 of file KLInferenceMethod.h.

float64_t m_noise_factor
protectedinherited

The factor used to ensure kernel matrix to be positive definite

Definition at line 291 of file KLInferenceMethod.h.

float64_t m_orthantwise_c
protectedinherited

Definition at line 464 of file KLInferenceMethod.h.

int m_orthantwise_end
protectedinherited

Definition at line 470 of file KLInferenceMethod.h.

int m_orthantwise_start
protectedinherited

Definition at line 467 of file KLInferenceMethod.h.

ParameterMap* m_parameter_map
inherited

map for different parameter versions

Definition at line 514 of file SGObject.h.

Parameter* m_parameters
inherited

parameters

Definition at line 505 of file SGObject.h.

int m_past
protectedinherited

Definition at line 440 of file KLInferenceMethod.h.

SGVector<float64_t> m_s2
protectedinherited

variational parameter sigma2 Note that sigma2 = diag(m_Sigma)

Definition at line 422 of file KLInferenceMethod.h.

float64_t m_scale
protectedinherited

kernel scale

Definition at line 449 of file InferenceMethod.h.

SGMatrix<float64_t> m_Sigma
protectedinherited

covariance matrix of the approximation to the posterior

Definition at line 417 of file KLInferenceMethod.h.

float64_t m_wolfe
protectedinherited

Definition at line 455 of file KLInferenceMethod.h.

float64_t m_xtol
protectedinherited

Definition at line 461 of file KLInferenceMethod.h.

Parallel* parallel
inherited

parallel

Definition at line 499 of file SGObject.h.

Version* version
inherited

version

Definition at line 502 of file SGObject.h.


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

SHOGUN Machine Learning Toolbox - Documentation