PLearn 0.1
Public Member Functions | Static Public Member Functions | Public Attributes | Static Public Attributes | Static Protected Member Functions | Private Types | Private Member Functions
PLearn::RandomGaussMix Class Reference

#include <RandomGaussMix.h>

Inheritance diagram for PLearn::RandomGaussMix:
Inheritance graph
[legend]
Collaboration diagram for PLearn::RandomGaussMix:
Collaboration graph
[legend]

List of all members.

Public Member Functions

 RandomGaussMix ()
 Default constructor.
virtual void train ()
 The role of the train method is to bring the learner up to stage == nstages, updating the train_stats collector with training costs measured on-line in the process.
virtual string classname () const
virtual OptionListgetOptionList () const
virtual OptionMapgetOptionMap () const
virtual RemoteMethodMapgetRemoteMethodMap () const
virtual RandomGaussMixdeepCopy (CopiesMap &copies) const
virtual void build ()
 Simply calls inherited::build() then build_().
virtual void makeDeepCopyFromShallowCopy (CopiesMap &copies)
 Transforms a shallow copy into a deep copy.

Static Public Member Functions

static string _classname_ ()
 Declares name and deepCopy methods.
static OptionList_getOptionList_ ()
static RemoteMethodMap_getRemoteMethodMap_ ()
static Object_new_instance_for_typemap_ ()
static bool _isa_ (const Object *o)
static void _static_initialize_ ()
static const PPathdeclaringFile ()

Public Attributes

PP< PDistributionmean_distribution
PP< PDistributionvariance_distribution
PP< PDistributionweight_distribution

Static Public Attributes

static StaticInitializer _static_initializer_

Static Protected Member Functions

static void declareOptions (OptionList &ol)
 Declares the class options.

Private Types

typedef GaussMix inherited

Private Member Functions

void build_ ()
 This does the actual building.

Detailed Description

Todo:
Synchronize with C++ help.

Definition at line 50 of file RandomGaussMix.h.


Member Typedef Documentation

Reimplemented from PLearn::GaussMix.

Definition at line 52 of file RandomGaussMix.h.


Constructor & Destructor Documentation

PLearn::RandomGaussMix::RandomGaussMix ( )

Default constructor.

Definition at line 61 of file RandomGaussMix.cc.

{
    type = "general";
}

Member Function Documentation

string PLearn::RandomGaussMix::_classname_ ( ) [static]

Declares name and deepCopy methods.

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

OptionList & PLearn::RandomGaussMix::_getOptionList_ ( ) [static]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

RemoteMethodMap & PLearn::RandomGaussMix::_getRemoteMethodMap_ ( ) [static]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

bool PLearn::RandomGaussMix::_isa_ ( const Object o) [static]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

Object * PLearn::RandomGaussMix::_new_instance_for_typemap_ ( ) [static]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

StaticInitializer RandomGaussMix::_static_initializer_ & PLearn::RandomGaussMix::_static_initialize_ ( ) [static]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

void PLearn::RandomGaussMix::build ( ) [virtual]

Simply calls inherited::build() then build_().

Reimplemented from PLearn::GaussMix.

Definition at line 192 of file RandomGaussMix.cc.

void PLearn::RandomGaussMix::build_ ( ) [private]

This does the actual building.

Reimplemented from PLearn::GaussMix.

Definition at line 201 of file RandomGaussMix.cc.

References PLearn::center(), PLearn::GramSchmidtOrthogonalization(), j, PLearn::TVec< T >::length(), PLearn::min(), PLASSERT, PLERROR, PLearn::TMat< T >::resize(), PLearn::sortElements(), PLearn::TVec< T >::subVec(), PLearn::sum(), and PLearn::TVec< T >::swap().

{
    if (!variance_distribution || !mean_distribution || !weight_distribution)
        return;

    // Need to reset the underlying distributions' seeds so that the generated
    // values are always the same at each build.
    mean_distribution->resetGenerator(mean_distribution->seed_);
    variance_distribution->resetGenerator(variance_distribution->seed_);
    weight_distribution->resetGenerator(weight_distribution->seed_);

    D = mean_distribution->getNPredicted();

    // Generate random Gaussian parameters.
    eigenvalues.resize(L, D);
    center.resize(L, D);
    alpha.resize(L);
    eigenvectors.resize(L);
    for (int j = 0; j < L; j++) {
        // Generate random matrix and perform Gram-Schmidt orthonormalization.
        Mat& eigenvecs = eigenvectors[j];
        eigenvecs.resize(D, D);
        int n_basis = -1;
        // It might happen that the rows of the random matrix are not
        // sufficiently independent, in which case we just try again.
        while (n_basis != D) {
            random_gen->fill_random_uniform(eigenvecs, -1, 1);
            n_basis = GramSchmidtOrthogonalization(eigenvecs);
        }
        // Generate random eigenvalues.
        Vec eigenvals = eigenvalues(j);
        variance_distribution->generate(eigenvals);
        PLASSERT( eigenvals.length() == D );
        // Note that eigenvalues must be sorted in decreasing order.
        sortElements(eigenvals);
        eigenvals.swap();
        // Generate random mean.
        Vec mean_j = center(j);
        mean_distribution->generate(mean_j);
        PLASSERT( mean_j.length() == D );
        // Generate random weight.
        Vec alpha_j = alpha.subVec(j, 1);
        weight_distribution->generate(alpha_j);
        PLASSERT( alpha_j.length() == 1 );
    }
    // Normalize 'alpha' so that it sums to 1.
    real sum = 0;
    for (int j = 0; j < L; j++) {
        real alpha_j = alpha[j];
        if (alpha_j < 0)
            PLERROR("In RandomGaussMix::build_ - The weight of a Gaussian "
                    "cannot be negative");
        sum += alpha_j;
    }
    PLASSERT( sum > 0 );
    alpha /= sum;
    // Set a few parameters that are needed.
    alpha_min = min(alpha);
    sigma_min = 1e-10;
    inputsize_ = D;
    n_eigen_computed = D;
    stage = 1;
    // Rebuild the mixture.
    inherited::build();
}

Here is the call graph for this function:

string PLearn::RandomGaussMix::classname ( ) const [virtual]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

void PLearn::RandomGaussMix::declareOptions ( OptionList ol) [static, protected]

Declares the class options.

Reimplemented from PLearn::GaussMix.

Definition at line 69 of file RandomGaussMix.cc.

References PLearn::GaussMix::alpha_min, PLearn::OptionBase::buildoption, PLearn::declareOption(), PLearn::GaussMix::efficient_k_median, PLearn::GaussMix::efficient_k_median_iter, PLearn::GaussMix::efficient_missing, PLearn::GaussMix::epsilon, PLearn::PLearner::expdir, PLearn::PLearner::forget_when_training_set_changes, PLearn::GaussMix::impute_missing, PLearn::GaussMix::kmeans_iterations, mean_distribution, PLearn::GaussMix::n_eigen, PLearn::PDistribution::n_predictor, PLearn::OptionBase::nosave, PLearn::PLearner::nservers, PLearn::PLearner::nstages, PLearn::PDistribution::predicted_size, PLearn::PDistribution::predictor_part, PLearn::PDistribution::predictor_size, PLearn::redeclareOption(), PLearn::PLearner::save_trainingset_prefix, PLearn::GaussMix::sigma_min, PLearn::GaussMix::type, variance_distribution, and weight_distribution.

{
    // ### Declare all of this object's options here
    // ### For the "flags" of each option, you should typically specify
    // ### one of OptionBase::buildoption, OptionBase::learntoption or
    // ### OptionBase::tuningoption. Another possible flag to be combined with
    // ### is OptionBase::nosave

    declareOption(ol, "mean_distribution",
                      &RandomGaussMix::mean_distribution,
                      OptionBase::buildoption,
        "The distribution from which means are sampled. A sample from this\n"
        "distribution should be a D-dimensional vector, representing the mean\n"
        "of a Gaussian in the input space.");

    declareOption(ol, "variance_distribution",
                      &RandomGaussMix::variance_distribution,
                      OptionBase::buildoption,
        "The distribution from which variances are sampled. A sample from\n"
        "this distribution should be a D-dimensional vector, representing\n"
        "the variance in each of the D principal directions of the Gaussian.");

    declareOption(ol, "weight_distribution",
                      &RandomGaussMix::weight_distribution,
                      OptionBase::buildoption,
        "The distribution from which the weight of each Gaussian is sampled.\n"
        "It should output a single non-negative scalar (the weights will be\n"
        "normalized afterwards so that they sum to 1).");

    // Now call the parent class' declareOptions().
    inherited::declareOptions(ol);

    // Hide unused options.
    redeclareOption(ol, "type", &RandomGaussMix::type,
                                OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "n_eigen", &RandomGaussMix::n_eigen,
                                   OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "efficient_missing",
                         &RandomGaussMix::efficient_missing,
                         OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "efficient_k_median",
                        &RandomGaussMix::efficient_k_median,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "efficient_k_median_iter",
                        &RandomGaussMix::efficient_k_median_iter,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "impute_missing", &RandomGaussMix::impute_missing,
                                          OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "kmeans_iterations",
                        &RandomGaussMix::kmeans_iterations,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "alpha_min", &RandomGaussMix::alpha_min,
                                     OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol,"sigma_min", &RandomGaussMix::sigma_min,
                                     OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "epsilon", &RandomGaussMix::epsilon,
                                   OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "predictor_size",
                        &RandomGaussMix::predictor_size,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "predicted_size",
                        &RandomGaussMix::predicted_size,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "predictor_part",
                        &RandomGaussMix::predictor_part,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "n_predictor",
                        &RandomGaussMix::n_predictor,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "expdir", &RandomGaussMix::expdir,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "forget_when_training_set_changes",
                        &RandomGaussMix::forget_when_training_set_changes,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "nstages", &RandomGaussMix::nstages,
                                   OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "nservers", &RandomGaussMix::nservers,
                                    OptionBase::nosave,
        "Not used in RandomGaussMix.");

    redeclareOption(ol, "save_trainingset_prefix",
                        &RandomGaussMix::save_trainingset_prefix,
                        OptionBase::nosave,
        "Not used in RandomGaussMix.");
}

Here is the call graph for this function:

static const PPath& PLearn::RandomGaussMix::declaringFile ( ) [inline, static]

Reimplemented from PLearn::GaussMix.

Definition at line 77 of file RandomGaussMix.h.

:
    //#####  Protected Options  ###############################################
RandomGaussMix * PLearn::RandomGaussMix::deepCopy ( CopiesMap copies) const [virtual]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

OptionList & PLearn::RandomGaussMix::getOptionList ( ) const [virtual]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

OptionMap & PLearn::RandomGaussMix::getOptionMap ( ) const [virtual]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

RemoteMethodMap & PLearn::RandomGaussMix::getRemoteMethodMap ( ) const [virtual]

Reimplemented from PLearn::GaussMix.

Definition at line 56 of file RandomGaussMix.cc.

void PLearn::RandomGaussMix::makeDeepCopyFromShallowCopy ( CopiesMap copies) [virtual]

Transforms a shallow copy into a deep copy.

Reimplemented from PLearn::GaussMix.

Definition at line 270 of file RandomGaussMix.cc.

References PLERROR.

{
    inherited::makeDeepCopyFromShallowCopy(copies);

    // ### Call deepCopyField on all "pointer-like" fields
    // ### that you wish to be deepCopied rather than
    // ### shallow-copied.
    // ### ex:
    // deepCopyField(trainvec, copies);

    // ### Remove this line when you have fully implemented this method.
    PLERROR("RandomGaussMix::makeDeepCopyFromShallowCopy not fully (correctly) implemented yet!");
}
void PLearn::RandomGaussMix::train ( ) [virtual]

The role of the train method is to bring the learner up to stage == nstages, updating the train_stats collector with training costs measured on-line in the process.

Reimplemented from PLearn::GaussMix.

Definition at line 287 of file RandomGaussMix.cc.

{
    // This class does not need to be trained.
    return;
}

Member Data Documentation

Reimplemented from PLearn::GaussMix.

Definition at line 77 of file RandomGaussMix.h.

Definition at line 57 of file RandomGaussMix.h.

Referenced by declareOptions().

Definition at line 58 of file RandomGaussMix.h.

Referenced by declareOptions().

Definition at line 59 of file RandomGaussMix.h.

Referenced by declareOptions().


The documentation for this class was generated from the following files:
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Defines