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

Does the same thing as Hinton's deep belief nets. More...

#include <GaussianDBNClassification.h>

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

List of all members.

Public Member Functions

 GaussianDBNClassification ()
 Default constructor.
virtual real density (const Vec &y) const
 Return probability density p(y | x)
virtual real log_density (const Vec &y) const
 Return log of probability density log(p(y | x)).
virtual real survival_fn (const Vec &y) const
 Return survival function: P(Y>y | x).
virtual real cdf (const Vec &y) const
 Return cdf: P(Y<y | x).
virtual void expectation (Vec &mu) const
 Return E[Y | x].
virtual void variance (Mat &cov) const
 Return Var[Y | x].
virtual void generate (Vec &y) const
 Return a pseudo-random sample generated from the conditional distribution, of density p(y | x).
virtual bool setPredictorPredictedSizes (int the_predictor_size, int the_predicted_size, bool call_parent=true)
 Generates a pseudo-random sample x from the reversed conditional distribution, of density p(x | y) (and NOT p(y | x)).
virtual void setPredictor (const Vec &predictor, bool call_parent=true) const
 Set the value for the predictor part of a conditional probability.
virtual void forget ()
 (Re-)initializes the PDistribution in its fresh state (that state may depend on the 'seed' option).
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 void computeCostsFromOutputs (const Vec &input, const Vec &output, const Vec &target, Vec &costs) const
 Compute a cost, depending on the type of the first output : if it is the density or the log-density: NLL if it is the expectation: NLL and class error.
virtual TVec< string > getTestCostNames () const
 Return [ "NLL" ] (the only cost computed by a PDistribution).
virtual string classname () const
virtual OptionListgetOptionList () const
virtual OptionMapgetOptionMap () const
virtual RemoteMethodMapgetRemoteMethodMap () const
virtual GaussianDBNClassificationdeepCopy (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_ ()
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

real learning_rate
 ### declare public option fields (such as build options) here Start your comments with Doxygen-compatible comments such as //!
real weight_decay
 The weight decay.
string initialization_method
 The method used to initialize the weights:
int n_layers
 Number of layers, including input layer and last layer, but not target layer.
TVec< PP< RBMLayer > > layers
 Layers that learn representations of the input, layers[0] is input layer, layers[n_layers-1] is last layer.
PP< RBMLayerlast_layer
 Last layer, learning joint representations of input and target.
PP< RBMMultinomialLayertarget_layer
 Target (or label) layer.
PP< RBMMixedLayerjoint_layer
 Concatenation of target_layer and layers[n_layers-2].
TVec< PP< RBMLLParameters > > params
 RBMParameters linking the unsupervised layers.
PP< RBMQLParametersinput_params
PP< RBMLLParameterstarget_params
 Parameters linking target_layer and last_layer.
PP< RBMJointLLParametersjoint_params
 Parameters linking joint_layer and last_layer.
TVec< inttraining_schedule
 Number of examples to use during each of the different greedy steps of the training phase.
string fine_tuning_method
 Method for fine-tuning the whole network after greedy learning.
bool use_sample_rather_than_expectation_in_positive_phase_statistics

Static Public Attributes

static StaticInitializer _static_initializer_

Protected Member Functions

virtual void greedyStep (const Vec &predictor, int params_index)
virtual void jointGreedyStep (const Vec &input)
virtual void fineTuneByGradientDescent (const Vec &input)

Static Protected Member Functions

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

Protected Attributes

TVec< Vecactivation_gradients
 gradients of cost wrt the activations (output of params)
TVec< Vecexpectation_gradients
 gradients of cost wrt the expectations (output of layers)
Vec output_gradient
 gradient wrt output activations

Private Types

typedef PDistribution inherited

Private Member Functions

void build_ ()
 This does the actual building.
void build_layers ()
 Build the layers.
void build_params ()
 Build the parameters if needed.

Detailed Description

Does the same thing as Hinton's deep belief nets.

Todo:
Yes
Deprecated:
Use ../DeepBeliefNet.h instead

Definition at line 67 of file GaussianDBNClassification.h.


Member Typedef Documentation

Reimplemented from PLearn::PDistribution.

Definition at line 69 of file GaussianDBNClassification.h.


Constructor & Destructor Documentation

PLearn::GaussianDBNClassification::GaussianDBNClassification ( )

Default constructor.

Definition at line 63 of file GaussianDBNClassification.cc.

References PLearn::PLearner::random_gen.


Member Function Documentation

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

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

Reimplemented from PLearn::PDistribution.

Definition at line 168 of file GaussianDBNClassification.cc.

References PLearn::PDistribution::build(), and build_().

{
    // ### Nothing to add here, simply calls build_().
    inherited::build();
    build_();
}

Here is the call graph for this function:

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

This does the actual building.

Reimplemented from PLearn::PDistribution.

Definition at line 178 of file GaussianDBNClassification.cc.

References build_layers(), build_params(), PLearn::endl(), fine_tuning_method, initialization_method, layers, PLearn::TVec< T >::length(), PLearn::lowerstring(), n_layers, PLERROR, and training_schedule.

Referenced by build().

{
    MODULE_LOG << "build_() called" << endl;
    n_layers = layers.length();
    if( n_layers <= 1 )
        return;

    // check value of initialization_method
    string im = lowerstring( initialization_method );
    if( im == "" || im == "uniform_sqrt" )
        initialization_method = "uniform_sqrt";
    else if( im == "uniform_linear" )
        initialization_method = im;
    else if( im == "zero" )
        initialization_method = im;
    else
        PLERROR( "RBMParameters::build_ - initialization_method\n"
                 "\"%s\" unknown.\n", initialization_method.c_str() );
    MODULE_LOG << "  initialization_method = \"" << initialization_method
        << "\"" << endl;

    // check value of fine_tuning_method
    string ftm = lowerstring( fine_tuning_method );
    if( ftm == "" | ftm == "none" )
        fine_tuning_method = "";
    else if( ftm == "cd" | ftm == "contrastive_divergence" )
        fine_tuning_method = "CD";
    else if( ftm == "egd" | ftm == "error_gradient_descent" )
        fine_tuning_method = "EGD";
    else if( ftm == "ws" | ftm == "wake_sleep" )
        fine_tuning_method = "WS";
    else
        PLERROR( "GaussianDBNClassification::build_ - fine_tuning_method \"%s\"\n"
                 "is unknown.\n", fine_tuning_method.c_str() );
    MODULE_LOG << "  fine_tuning_method = \"" << fine_tuning_method << "\""
        <<  endl;
    //TODO: build structure to store gradients during gradient descent

    if( training_schedule.length() != n_layers )
        training_schedule = TVec<int>( n_layers, 1000000 );
    MODULE_LOG << "  training_schedule = " << training_schedule << endl;
    MODULE_LOG << endl;

    build_layers();
    build_params();
}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::build_layers ( ) [private]

Build the layers.

Definition at line 225 of file GaussianDBNClassification.cc.

References PLearn::endl(), i, PLearn::PLearner::inputsize(), PLearn::PLearner::inputsize_, joint_layer, last_layer, layers, n_layers, PLearn::PDistribution::n_predicted, PLearn::PDistribution::n_predictor, PLASSERT, PLearn::PLearner::random_gen, setPredictorPredictedSizes(), and target_layer.

Referenced by build_().

{
    MODULE_LOG << "build_layers() called" << endl;
    if( inputsize_ >= 0 )
    {
        PLASSERT( layers[0]->size + target_layer->size == inputsize() );
        setPredictorPredictedSizes( layers[0]->size,
                                    target_layer->size, false );
        MODULE_LOG << "  n_predictor = " << n_predictor << endl;
        MODULE_LOG << "  n_predicted = " << n_predicted << endl;
    }

    for( int i=0 ; i<n_layers ; i++ )
        layers[i]->random_gen = random_gen;
    target_layer->random_gen = random_gen;

    last_layer = layers[n_layers-1];

    // concatenate target_layer and layers[n_layers-2] into joint_layer
    TVec< PP<RBMLayer> > the_sub_layers( 2 );
    the_sub_layers[0] = target_layer;
    the_sub_layers[1] = layers[n_layers-2];
    joint_layer = new RBMMixedLayer( the_sub_layers );
    joint_layer->random_gen = random_gen;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::build_params ( ) [private]

Build the parameters if needed.

Definition at line 251 of file GaussianDBNClassification.cc.

References activation_gradients, PLearn::endl(), expectation_gradients, i, initialization_method, input_params, joint_params, last_layer, layers, learning_rate, PLearn::TVec< T >::length(), n_layers, PLearn::PDistribution::n_predicted, output_gradient, params, PLERROR, PLearn::PLearner::random_gen, PLearn::TVec< T >::resize(), target_layer, and target_params.

Referenced by build_().

{
    MODULE_LOG << "build_params() called" << endl;
    if( params.length() == 0 )
    {
        input_params = new RBMQLParameters() ; 
        params.resize( n_layers-1 );
        for( int i=1 ; i<n_layers-1 ; i++ )
            params[i] = new RBMLLParameters();
        // params[0] is not being using, it is not being created
    }
    else if( params.length() != n_layers-1 )
        PLERROR( "GaussianDBNClassification::build_params - params.length() should\n"
                 "be equal to layers.length()-1 (%d != %d).\n",
                 params.length(), n_layers-1 );

    activation_gradients.resize( n_layers-1 );
    expectation_gradients.resize( n_layers-1 );
    output_gradient.resize( n_predicted );

    input_params->down_units_types = layers[0]->units_types;
    input_params->up_units_types = layers[1]->units_types;
    input_params->learning_rate = learning_rate;
    input_params->initialization_method = initialization_method;
    input_params->random_gen = random_gen;
    input_params->build();

    activation_gradients[0].resize( input_params->down_layer_size );
    expectation_gradients[0].resize( input_params->down_layer_size );


    for( int i=1 ; i<n_layers-1 ; i++ )
    {
        //TODO: call changeOptions instead
        
        params[i]->down_units_types = layers[i]->units_types;
        params[i]->up_units_types = layers[i+1]->units_types;
        params[i]->learning_rate = learning_rate;
        params[i]->initialization_method = initialization_method;
        params[i]->random_gen = random_gen;
        params[i]->build();

        activation_gradients[i].resize( params[i]->down_layer_size );
        expectation_gradients[i].resize( params[i]->down_layer_size );
    }

    if( target_layer && !target_params )
        target_params = new RBMLLParameters();

    //TODO: call changeOptions instead
    target_params->down_units_types = target_layer->units_types;
    target_params->up_units_types = last_layer->units_types;
    target_params->learning_rate = learning_rate;
    target_params->initialization_method = initialization_method;
    target_params->random_gen = random_gen;
    target_params->build();

    // build joint_params from params[n_layers-1] and target_params
    joint_params = new RBMJointLLParameters( target_params,
                                             params[n_layers-2] );
    joint_params->learning_rate = learning_rate;
    joint_params->random_gen = random_gen;
}

Here is the call graph for this function:

Here is the caller graph for this function:

real PLearn::GaussianDBNClassification::cdf ( const Vec y) const [virtual]

Return cdf: P(Y<y | x).

Reimplemented from PLearn::PDistribution.

Definition at line 352 of file GaussianDBNClassification.cc.

References PLERROR.

{
    PLERROR("cdf not implemented for GaussianDBNClassification"); return 0;
}
string PLearn::GaussianDBNClassification::classname ( ) const [virtual]

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

Referenced by train().

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::computeCostsFromOutputs ( const Vec input,
const Vec output,
const Vec target,
Vec costs 
) const [virtual]

Compute a cost, depending on the type of the first output : if it is the density or the log-density: NLL if it is the expectation: NLL and class error.

Reimplemented from PLearn::PDistribution.

Definition at line 827 of file GaussianDBNClassification.cc.

References PLearn::argmax(), c, PLearn::PDistribution::computeCostsFromOutputs(), i, PLearn::is_equal(), PLearn::PDistribution::n_predicted, PLearn::PDistribution::outputs_def, pl_log, PLASSERT, PLearn::PDistribution::predicted_part, PLearn::TVec< T >::resize(), and PLearn::PDistribution::splitCond().

{
    char c = outputs_def[0];
    if( c == 'l' || c == 'd' )
        inherited::computeCostsFromOutputs(input, output, target, costs);
    else if( c == 'e' )
    {
        costs.resize( 2 );
        splitCond(input);

        // actual_index is the actual 'target'
        int actual_index = argmax(predicted_part);
#ifdef BOUNDCHECK
        for( int i=0 ; i<n_predicted ; i++ )
            PLASSERT( is_equal( predicted_part[i], 0. ) ||
                    i == actual_index && is_equal( predicted_part[i], 1. ) );
#endif
        costs[0] = -pl_log( output[actual_index] );

        // predicted_index is the most probable predicted class
        int predicted_index = argmax(output);
        if( predicted_index == actual_index )
            costs[1] = 0;
        else
            costs[1] = 1;
    }
}

Here is the call graph for this function:

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

Declares the class options.

Reimplemented from PLearn::PDistribution.

Definition at line 74 of file GaussianDBNClassification.cc.

References PLearn::OptionBase::buildoption, PLearn::declareOption(), PLearn::PDistribution::declareOptions(), fine_tuning_method, initialization_method, input_params, joint_layer, joint_params, last_layer, layers, learning_rate, PLearn::OptionBase::learntoption, n_layers, params, target_layer, target_params, training_schedule, use_sample_rather_than_expectation_in_positive_phase_statistics, and weight_decay.

{
    declareOption(ol, "learning_rate", &GaussianDBNClassification::learning_rate,
                  OptionBase::buildoption,
                  "Learning rate");

    declareOption(ol, "weight_decay", &GaussianDBNClassification::weight_decay,
                  OptionBase::buildoption,
                  "Weight decay");

    declareOption(ol, "initialization_method",
                  &GaussianDBNClassification::initialization_method,
                  OptionBase::buildoption,
                  "The method used to initialize the weights:\n"
                  "  - \"uniform_linear\" = a uniform law in [-1/d, 1/d]\n"
                  "  - \"uniform_sqrt\"   = a uniform law in [-1/sqrt(d),"
                  " 1/sqrt(d)]\n"
                  "  - \"zero\"           = all weights are set to 0,\n"
                  "where d = max( up_layer_size, down_layer_size ).\n");


    declareOption(ol, "training_schedule",
                  &GaussianDBNClassification::training_schedule,
                  OptionBase::buildoption,
                  "Number of examples to use during each of the different"
                  " greedy\n"
                  "steps of the training phase.\n");

    declareOption(ol, "fine_tuning_method",
                  &GaussianDBNClassification::fine_tuning_method,
                  OptionBase::buildoption,
                  "Method for fine-tuning the whole network after greedy"
                  " learning.\n"
                  "One of:\n"
                  "  - \"none\"\n"
                  "  - \"CD\" or \"contrastive_divergence\"\n"
                  "  - \"EGD\" or \"error_gradient_descent\"\n"
                  "  - \"WS\" or \"wake_sleep\".\n");

    declareOption(ol, "layers", &GaussianDBNClassification::layers,
                  OptionBase::buildoption,
                  "Layers that learn representations of the input,"
                  " unsupervisedly.\n"
                  "layers[0] is input layer.\n");

    declareOption(ol, "target_layer", &GaussianDBNClassification::target_layer,
                  OptionBase::buildoption,
                  "Target (or label) layer");

    declareOption(ol, "params", &GaussianDBNClassification::params,
                  OptionBase::buildoption,
                  "RBMParameters linking the unsupervised layers.\n"
                  "params[i] links layers[i] and layers[i+1], except for"
                  "params[n_layers-1],\n"
                  "that links layers[n_layers-1] and last_layer.\n");

    declareOption(ol, "target_params", &GaussianDBNClassification::target_params,
                  OptionBase::buildoption,
                  "Parameters linking target_layer and last_layer");
    
    declareOption(ol, "input_params", &GaussianDBNClassification::input_params,
                  OptionBase::buildoption,
                  "Parameters linking layer[0] and layer[1]");

    declareOption(ol, "use_sample_rather_than_expectation_in_positive_phase_statistics",
                  &GaussianDBNClassification::use_sample_rather_than_expectation_in_positive_phase_statistics,
                  OptionBase::buildoption,
                  "In positive phase statistics use output->sample * input\n"
                  "rather than output->expectation * input.\n");

    declareOption(ol, "n_layers", &GaussianDBNClassification::n_layers,
                  OptionBase::learntoption,
                  "Number of unsupervised layers, including input layer");

    declareOption(ol, "last_layer", &GaussianDBNClassification::last_layer,
                  OptionBase::learntoption,
                  "Last layer, learning joint representations of input and"
                  " target");

    declareOption(ol, "joint_layer", &GaussianDBNClassification::joint_layer,
                  OptionBase::learntoption,
                  "Concatenation of target_layer and layers[n_layers-1]");

    declareOption(ol, "joint_params", &GaussianDBNClassification::joint_params,
                  OptionBase::learntoption,
                  "Parameters linking joint_layer and last_layer");

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

Here is the call graph for this function:

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

Reimplemented from PLearn::PDistribution.

Definition at line 245 of file GaussianDBNClassification.h.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

real PLearn::GaussianDBNClassification::density ( const Vec y) const [virtual]

Return probability density p(y | x)

Reimplemented from PLearn::PDistribution.

Definition at line 391 of file GaussianDBNClassification.cc.

References PLearn::argmax(), expectation(), i, PLearn::is_equal(), PLearn::PDistribution::n_predicted, PLASSERT, PLearn::TVec< T >::size(), and PLearn::PDistribution::store_expect.

Referenced by log_density().

{
    PLASSERT( y.size() == n_predicted );

    // TODO: 'y'[0] devrait plutot etre l'entier "index" lui-meme!
    int index = argmax( y );

    // If y != onehot( index ), then density is 0
    if( !is_equal( y[index], 1. ) )
        return 0;
    for( int i=0 ; i<n_predicted ; i++ )
        if( !is_equal( y[i], 0 ) && i != index )
            return 0;

    expectation( store_expect );
    return store_expect[index];
}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::expectation ( Vec mu) const [virtual]

Return E[Y | x].

Reimplemented from PLearn::PDistribution.

Definition at line 360 of file GaussianDBNClassification.cc.

References i, input_params, joint_params, layers, n_layers, params, PLearn::PDistribution::predicted_size, PLearn::PDistribution::predictor_part, PLearn::TVec< T >::resize(), and target_layer.

Referenced by density(), fineTuneByGradientDescent(), greedyStep(), and jointGreedyStep().

{
    mu.resize( predicted_size );

    // Propagate input (predictor_part) until penultimate layer
    layers[0]->expectation << predictor_part;
    input_params->setAsDownInput(layers[0]->expectation) ; 
    layers[1]->getAllActivations( (RBMQLParameters*) input_params );
    layers[1]->computeExpectation();
    
    for( int i=1 ; i<n_layers-2 ; i++ )
    {
        params[i]->setAsDownInput( layers[i]->expectation );
        layers[i+1]->getAllActivations( (RBMLLParameters*) params[i] );
        layers[i+1]->computeExpectation();
    }

    // Set layers[n_layers-2]->expectation (penultimate) as conditionning input
    // of joint_params
    joint_params->setAsCondInput( layers[n_layers-2]->expectation );

    // Get all activations on target_layer from target_params
    target_layer->getAllActivations( (RBMLLParameters*) joint_params );
    target_layer->computeExpectation();

    mu << target_layer->expectation;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::fineTuneByGradientDescent ( const Vec input) [protected, virtual]

Definition at line 797 of file GaussianDBNClassification.cc.

References activation_gradients, PLearn::argmax(), expectation(), expectation_gradients, i, joint_params, layers, n_layers, output_gradient, params, PLearn::PDistribution::predicted_part, PLearn::PDistribution::splitCond(), and target_layer.

Referenced by train().

{
    // split input in predictor_part and predicted_part
    splitCond(input);

    // compute predicted_part expectation, conditioned on predictor_part
    // (forward pass)
    expectation( output_gradient );

    int actual_index = argmax(predicted_part);
    output_gradient[actual_index] -= 1.;

    joint_params->bpropUpdate( layers[n_layers-2]->expectation,
                               target_layer->expectation,
                               expectation_gradients[n_layers-2],
                               output_gradient );

    for( int i=n_layers-2 ; i>0 ; i-- )
    {
        layers[i]->bpropUpdate( layers[i]->activations,
                                layers[i]->expectation,
                                activation_gradients[i],
                                expectation_gradients[i] );
        params[i-1]->bpropUpdate( layers[i-1]->expectation,
                                  layers[i]->activations,
                                  expectation_gradients[i-1],
                                  activation_gradients[i] );
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::forget ( ) [virtual]

(Re-)initializes the PDistribution in its fresh state (that state may depend on the 'seed' option).

And sets 'stage' back to 0 (this is the stage of a fresh learner!). ### You may remove this method if your distribution does not ### implement it.

A typical forget() method should do the following:

  • initialize a random number generator with the seed option
  • initialize the learner's parameters, using this random generator
  • stage = 0

Reimplemented from PLearn::PDistribution.

Definition at line 318 of file GaussianDBNClassification.cc.

References PLearn::endl(), i, input_params, layers, n_layers, params, PLearn::PDistribution::resetGenerator(), PLearn::PLearner::seed_, PLearn::PLearner::stage, target_layer, and target_params.

{
    MODULE_LOG << "forget() called" << endl;
    resetGenerator(seed_);
    input_params->forget() ; 
    for( int i=1 ; i<n_layers-1 ; i++ )
        params[i]->forget();

    for( int i=0 ; i<n_layers ; i++ )
        layers[i]->reset();

    target_params->forget();
    target_layer->reset();

    stage = 0;
}

Here is the call graph for this function:

void PLearn::GaussianDBNClassification::generate ( Vec y) const [virtual]

Return a pseudo-random sample generated from the conditional distribution, of density p(y | x).

Reimplemented from PLearn::PDistribution.

Definition at line 344 of file GaussianDBNClassification.cc.

References PLERROR.

{
    PLERROR("generate not implemented for GaussianDBNClassification");
}
OptionList & PLearn::GaussianDBNClassification::getOptionList ( ) const [virtual]

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

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

Reimplemented from PLearn::PDistribution.

Definition at line 58 of file GaussianDBNClassification.cc.

TVec< string > PLearn::GaussianDBNClassification::getTestCostNames ( ) const [virtual]

Return [ "NLL" ] (the only cost computed by a PDistribution).

Reimplemented from PLearn::PDistribution.

Definition at line 858 of file GaussianDBNClassification.cc.

References PLearn::TVec< T >::append(), c, and PLearn::PDistribution::outputs_def.

{
    char c = outputs_def[0];
    TVec<string> result;
    if( c == 'l' || c == 'd' )
        result.append( "NLL" );
    else if( c == 'e' )
    {
        result.append( "NLL" );
        result.append( "class_error" );
    }
    return result;
}

Here is the call graph for this function:

void PLearn::GaussianDBNClassification::greedyStep ( const Vec predictor,
int  params_index 
) [protected, virtual]

Definition at line 672 of file GaussianDBNClassification.cc.

References PLearn::RBMQLParameters::accumulateNegStats(), PLearn::RBMQLParameters::accumulatePosStats(), expectation(), i, input_params, layers, params, PLearn::sample(), PLearn::RBMParameters::setAsDownInput(), PLearn::RBMParameters::setAsUpInput(), PLearn::RBMQLParameters::update(), and use_sample_rather_than_expectation_in_positive_phase_statistics.

Referenced by train().

{
    // deterministic propagation until we reach index
    layers[0]->expectation << predictor;

    input_params->setAsDownInput( layers[0]->expectation );
    layers[1]->getAllActivations( (RBMQLParameters*) input_params );
    layers[1]->computeExpectation();
        
    for( int i=1 ; i<index ; i++ )
    {
        params[i]->setAsDownInput( layers[i]->expectation );
        layers[i+1]->getAllActivations( (RBMLLParameters*) params[i] );
        layers[i+1]->computeExpectation();
    }

    // positive phase
    if (index == 0) {
        input_params->setAsDownInput( layers[index]->expectation );
        layers[index+1]->getAllActivations((RBMQLParameters*) input_params);
        layers[index+1]->computeExpectation();
        layers[index+1]->generateSample();
        if (use_sample_rather_than_expectation_in_positive_phase_statistics)
            input_params->accumulatePosStats(layers[index]->expectation,
                    layers[index+1]->sample );
        else
            input_params->accumulatePosStats(layers[index]->expectation,
                    layers[index+1]->expectation );

        // down propagation
        input_params->setAsUpInput( layers[index+1]->sample );
        layers[index]->getAllActivations( (RBMQLParameters*) input_params );

        // negative phase
        layers[index]->generateSample();
        input_params->setAsDownInput( layers[index]->sample );
        layers[index+1]->getAllActivations((RBMQLParameters*) input_params);
        layers[index+1]->computeExpectation();
        input_params->accumulateNegStats( layers[index]->sample,
                layers[index+1]->expectation );

        // update
        input_params->update();

    }
    else {
        params[index]->setAsDownInput( layers[index]->expectation );
        layers[index+1]->getAllActivations((RBMLLParameters*) params[index]);
        layers[index+1]->computeExpectation();
        layers[index+1]->generateSample();
        if (use_sample_rather_than_expectation_in_positive_phase_statistics)
            params[index]->accumulatePosStats(layers[index]->expectation,
                    layers[index+1]->sample );
        else
            params[index]->accumulatePosStats(layers[index]->expectation,
                    layers[index+1]->expectation );

        // down propagation
        params[index]->setAsUpInput( layers[index+1]->sample );
        layers[index]->getAllActivations( (RBMLLParameters*) params[index] );

        // negative phase
        layers[index]->generateSample();
        params[index]->setAsDownInput( layers[index]->sample );
        layers[index+1]->getAllActivations((RBMLLParameters*) params[index]);
        layers[index+1]->computeExpectation();
        params[index]->accumulateNegStats( layers[index]->sample,
                layers[index+1]->expectation );

        // update
        params[index]->update();

    }
    

}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::GaussianDBNClassification::jointGreedyStep ( const Vec input) [protected, virtual]

Definition at line 749 of file GaussianDBNClassification.cc.

References PLearn::RBMLLParameters::accumulateNegStats(), PLearn::RBMLLParameters::accumulatePosStats(), expectation(), i, input_params, joint_layer, joint_params, last_layer, layers, n_layers, PLearn::PDistribution::n_predicted, PLearn::PDistribution::n_predictor, params, PLearn::RBMParameters::setAsDownInput(), PLearn::RBMParameters::setAsUpInput(), PLearn::TVec< T >::subVec(), target_layer, PLearn::RBMLLParameters::update(), and use_sample_rather_than_expectation_in_positive_phase_statistics.

Referenced by train().

{
    // deterministic propagation until we reach n_layers-2, setting the input
    // of the "input" part of joint_layer
    layers[0]->expectation << input.subVec( 0, n_predictor );
    input_params->setAsDownInput( layers[0]->expectation );
    layers[1]->getAllActivations( (RBMQLParameters*) input_params );
    layers[1]->computeExpectation();
    
    
    for( int i=1 ; i<n_layers-2 ; i++ )
    {
        params[i]->setAsDownInput( layers[i]->expectation );
        layers[i+1]->getAllActivations( (RBMLLParameters*) params[i] );
        layers[i+1]->computeExpectation();
    }

    // now fill the "target" part of joint_layer
    target_layer->expectation << input.subVec( n_predictor, n_predicted );

    // positive phase
    joint_params->setAsDownInput( joint_layer->expectation );
    last_layer->getAllActivations( (RBMLLParameters*) joint_params );
    last_layer->computeExpectation();
    last_layer->generateSample();
    if (use_sample_rather_than_expectation_in_positive_phase_statistics)
        joint_params->accumulatePosStats( joint_layer->expectation,
                                          last_layer->sample );
    else
        joint_params->accumulatePosStats( joint_layer->expectation,
                                          last_layer->expectation );

    // down propagation
    joint_params->setAsUpInput( last_layer->sample );
    joint_layer->getAllActivations( (RBMLLParameters*) joint_params );

    // negative phase
    joint_layer->generateSample();
    joint_params->setAsDownInput( joint_layer->sample );
    last_layer->getAllActivations( (RBMLLParameters*) joint_params );
    last_layer->computeExpectation();
    joint_params->accumulateNegStats( joint_layer->sample,
                                      last_layer->expectation );

    // update
    joint_params->update();
}

Here is the call graph for this function:

Here is the caller graph for this function:

real PLearn::GaussianDBNClassification::log_density ( const Vec y) const [virtual]

Return log of probability density log(p(y | x)).

Reimplemented from PLearn::PDistribution.

Definition at line 413 of file GaussianDBNClassification.cc.

References density(), and pl_log.

{
    return pl_log( density(y) );
}

Here is the call graph for this function:

void PLearn::GaussianDBNClassification::makeDeepCopyFromShallowCopy ( CopiesMap copies) [virtual]
void PLearn::GaussianDBNClassification::setPredictor ( const Vec predictor,
bool  call_parent = true 
) const [virtual]

Set the value for the predictor part of a conditional probability.

Reimplemented from PLearn::PDistribution.

Definition at line 455 of file GaussianDBNClassification.cc.

References PLearn::PDistribution::setPredictor().

{
    if (call_parent)
        inherited::setPredictor(predictor, true);
    // ### Add here any specific code required by your subclass.
}

Here is the call graph for this function:

bool PLearn::GaussianDBNClassification::setPredictorPredictedSizes ( int  the_predictor_size,
int  the_predicted_size,
bool  call_parent = true 
) [virtual]

Generates a pseudo-random sample x from the reversed conditional distribution, of density p(x | y) (and NOT p(y | x)).

i.e., generates a "predictor" part given a "predicted" part, regardless of any previously set predictor. Set the 'predictor' and 'predicted' sizes for this distribution.

Reimplemented from PLearn::PDistribution.

Definition at line 466 of file GaussianDBNClassification.cc.

References layers, PLearn::PDistribution::n_predicted, PLearn::PDistribution::n_predictor, PLERROR, PLearn::PDistribution::setPredictorPredictedSizes(), PLearn::TVec< T >::size(), and target_layer.

Referenced by build_layers().

{
    bool sizes_have_changed = false;
    if (call_parent)
        sizes_have_changed = inherited::setPredictorPredictedSizes(
            the_predictor_size, the_predicted_size, true);

    // ### Add here any specific code required by your subclass.
    if( the_predictor_size >= 0 && the_predictor_size != layers[0]->size ||
        the_predicted_size >= 0 && the_predicted_size != target_layer->size )
        PLERROR( "GaussianDBNClassification::setPredictorPredictedSizes - \n"
                 "n_predictor should be equal to layer[0]->size (%d)\n"
                 "n_predicted should be equal to target_layer->size (%d).\n",
                 layers[0]->size, target_layer->size );

    n_predictor = layers[0]->size;
    n_predicted = target_layer->size;

    // Returned value.
    return sizes_have_changed;
}

Here is the call graph for this function:

Here is the caller graph for this function:

real PLearn::GaussianDBNClassification::survival_fn ( const Vec y) const [virtual]

Return survival function: P(Y>y | x).

Reimplemented from PLearn::PDistribution.

Definition at line 421 of file GaussianDBNClassification.cc.

References PLERROR.

{
    PLERROR("survival_fn not implemented for GaussianDBNClassification"); return 0;
}
void PLearn::GaussianDBNClassification::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::PDistribution.

Definition at line 494 of file GaussianDBNClassification.cc.

References classname(), PLearn::endl(), fine_tuning_method, fineTuneByGradientDescent(), PLearn::VMat::getExample(), greedyStep(), i, PLearn::PLearner::initTrain(), PLearn::PLearner::inputsize(), jointGreedyStep(), PLearn::VMat::length(), n_layers, PLearn::PDistribution::n_predictor, PLearn::PLearner::nstages, PLERROR, PLearn::PLearner::report_progress, PLearn::sample(), PLearn::PLearner::stage, PLearn::TVec< T >::subVec(), PLearn::PLearner::targetsize(), PLearn::tostring(), PLearn::PLearner::train_set, PLearn::PLearner::train_stats, training_schedule, and PLearn::ProgressBar::update().

{
    MODULE_LOG << "train() called" << endl;
    // The role of the train method is to bring the learner up to
    // stage==nstages, updating train_stats with training costs measured
    // on-line in the process.

    /* TYPICAL CODE:

    static Vec input;  // static so we don't reallocate memory each time...
    static Vec target; // (but be careful that static means shared!)
    input.resize(inputsize());    // the train_set's inputsize()
    target.resize(targetsize());  // the train_set's targetsize()
    real weight;

    // This generic PLearner method does a number of standard stuff useful for
    // (almost) any learner, and return 'false' if no training should take
    // place. See PLearner.h for more details.
    if (!initTrain())
        return;

    while(stage<nstages)
    {
        // clear statistics of previous epoch
        train_stats->forget();

        //... train for 1 stage, and update train_stats,
        // using train_set->getExample(input, target, weight)
        // and train_stats->update(train_costs)

        ++stage;
        train_stats->finalize(); // finalize statistics for this epoch
    }
    */

    Vec input( inputsize() );
    Vec target( targetsize() ); // unused
    real weight; // unused

    if( !initTrain() )
    {
        MODULE_LOG << "train() aborted" << endl;
        return;
    }

    int nsamples = train_set->length();
    int sample = 0;
    MODULE_LOG << "  nsamples = " << nsamples << endl;

    // Let's define stage and nstages:
    //   - 0: fresh state, nothing is done
    //   - 1..n_layers-2: params[stage-1] is trained
    //   - n_layers-1: joint_params is trained (including params[n_layers-2])
    //   - n_layers: after the fine tuning

    MODULE_LOG << "initial stage = " << stage << endl;
    MODULE_LOG << "objective: nstages = " << nstages << endl;

    for( ; stage < nstages ; stage++ )
    {
        // clear stats of previous epoch
        train_stats->forget();

        // loops over the training set, until training_schedule[stage] examples
        // have been seen.
        // TODO: modify the training set used?
        int layer = stage;
        int n_samples_to_see = training_schedule[stage];

        // this progress bar shows the number of loops through the whole
        // training set
        ProgressBar* pb = 0;

        if( stage < n_layers-2 )
        {
            MODULE_LOG << "Training parameters between layers " << stage
                << " and " << stage+1 << endl;

            if( report_progress )
                pb = new ProgressBar( "Training " + classname()
                                      + " parameters between layers "
                                      + tostring(stage) + " and "
                                      + tostring(stage+1),
                                      n_samples_to_see );

            int begin_sample = sample;
            int end_sample = begin_sample + n_samples_to_see;
            for( ; sample < end_sample ; sample++ )
            {
                // sample is the index in the training set
                int i = sample % train_set->length();
                train_set->getExample(i, input, target, weight);
                greedyStep( input.subVec(0, n_predictor), layer );

                if( pb )
                    pb->update( sample - begin_sample + 1 );
            }

        }
        else if( stage == n_layers-2 )
        {
            MODULE_LOG << "Training joint parameters, between target,"
                << " penultimate (" << n_layers-2 << ")," << endl
                << "and last (" << n_layers-1 << ") layers." << endl;
            if( report_progress )
                pb = new ProgressBar( "Training " + classname()
                                      + " parameters between target, "
                                      + tostring(stage) + " and "
                                      + tostring(stage+1) + " layers",
                                      n_samples_to_see );

            int begin_sample = sample;
            int end_sample = begin_sample + n_samples_to_see;

            for( ; sample < end_sample ; sample++ )
            {
                // sample is the index in the training set
                int i = sample % train_set->length();
                train_set->getExample(i, input, target, weight);
                jointGreedyStep( input );

                if( pb )
                    pb->update( sample - begin_sample + 1 );
            }
        }
        else if( stage == n_layers-1 )
        {
            MODULE_LOG << "Fine-tuning all parameters, using method "
                << fine_tuning_method << endl;

            if( fine_tuning_method == "" ) // do nothing
                sample += n_samples_to_see;
            else if( fine_tuning_method == "EGD" )
            {
                if( report_progress )
                    pb = new ProgressBar( "Training all " + classname()
                                          + " parameters by fine tuning",
                                          n_samples_to_see );

/*
pout << "==================" << endl
    << "Before update:" << endl
    << "up:      " << joint_params->up_units_params << endl
    << "weights: " << endl << joint_params->weights << endl
    << "down:    " << joint_params->down_units_params << endl
    << endl;
// */
                int begin_sample = sample;
                int end_sample = begin_sample + n_samples_to_see;
                for( ; sample < end_sample ; sample++ )
                {
                    // sample is the index in the training set
                    int i = sample % train_set->length();
                    train_set->getExample(i, input, target, weight);
                    fineTuneByGradientDescent( input );

                    if( pb )
                        pb->update( sample - begin_sample + 1 );
                }
/*
pout << "-------" << endl
    << "After update:" << endl
    << "up:      " << joint_params->up_units_params << endl
    << "weights: " << endl << joint_params->weights << endl
    << "down:    " << joint_params->down_units_params << endl
    << endl;
// */
            }
            else
                PLERROR( "Fine-tuning methods other than \"EGD\" are not"
                         " implemented yet." );

        }
        train_stats->finalize(); // finalize statistics for this epoch
    }
    MODULE_LOG << endl;
}

Here is the call graph for this function:

void PLearn::GaussianDBNClassification::variance ( Mat cov) const [virtual]

Return Var[Y | x].

Reimplemented from PLearn::PDistribution.

Definition at line 429 of file GaussianDBNClassification.cc.

References PLERROR.

{
    PLERROR("variance not implemented for GaussianDBNClassification");
}

Member Data Documentation

Reimplemented from PLearn::PDistribution.

Definition at line 245 of file GaussianDBNClassification.h.

gradients of cost wrt the activations (output of params)

Definition at line 261 of file GaussianDBNClassification.h.

Referenced by build_params(), and fineTuneByGradientDescent().

gradients of cost wrt the expectations (output of layers)

Definition at line 264 of file GaussianDBNClassification.h.

Referenced by build_params(), and fineTuneByGradientDescent().

Method for fine-tuning the whole network after greedy learning.

One of:

  • "none"
  • "CD" or "contrastive_divergence"
  • "EGD" or "error_gradient_descent"
  • "WS" or "wake_sleep"

Definition at line 130 of file GaussianDBNClassification.h.

Referenced by build_(), declareOptions(), and train().

The method used to initialize the weights:

  • "uniform_linear" = a uniform law in [-1/d, 1/d]
  • "uniform_sqrt" = a uniform law in [-1/sqrt(d), 1/sqrt(d)]
  • "zero" = all weights are set to 0 Where d = max( up_layer_size, down_layer_size )

Definition at line 88 of file GaussianDBNClassification.h.

Referenced by build_(), build_params(), and declareOptions().

Concatenation of target_layer and layers[n_layers-2].

Definition at line 105 of file GaussianDBNClassification.h.

Referenced by build_layers(), declareOptions(), jointGreedyStep(), and makeDeepCopyFromShallowCopy().

Parameters linking joint_layer and last_layer.

Contains params[n_layers-2] and target_params.

Definition at line 118 of file GaussianDBNClassification.h.

Referenced by build_params(), declareOptions(), expectation(), fineTuneByGradientDescent(), jointGreedyStep(), and makeDeepCopyFromShallowCopy().

Last layer, learning joint representations of input and target.

Definition at line 99 of file GaussianDBNClassification.h.

Referenced by build_layers(), build_params(), declareOptions(), jointGreedyStep(), and makeDeepCopyFromShallowCopy().

Layers that learn representations of the input, layers[0] is input layer, layers[n_layers-1] is last layer.

Definition at line 96 of file GaussianDBNClassification.h.

Referenced by build_(), build_layers(), build_params(), declareOptions(), expectation(), fineTuneByGradientDescent(), forget(), greedyStep(), jointGreedyStep(), makeDeepCopyFromShallowCopy(), and setPredictorPredictedSizes().

### declare public option fields (such as build options) here Start your comments with Doxygen-compatible comments such as //!

The learning rate

Definition at line 78 of file GaussianDBNClassification.h.

Referenced by build_params(), and declareOptions().

Number of layers, including input layer and last layer, but not target layer.

Definition at line 92 of file GaussianDBNClassification.h.

Referenced by build_(), build_layers(), build_params(), declareOptions(), expectation(), fineTuneByGradientDescent(), forget(), jointGreedyStep(), and train().

gradient wrt output activations

Definition at line 267 of file GaussianDBNClassification.h.

Referenced by build_params(), and fineTuneByGradientDescent().

RBMParameters linking the unsupervised layers.

params[i] links layers[i] and layers[i+1], i>0

Definition at line 109 of file GaussianDBNClassification.h.

Referenced by build_params(), declareOptions(), expectation(), fineTuneByGradientDescent(), forget(), greedyStep(), jointGreedyStep(), and makeDeepCopyFromShallowCopy().

Parameters linking target_layer and last_layer.

Definition at line 114 of file GaussianDBNClassification.h.

Referenced by build_params(), declareOptions(), forget(), and makeDeepCopyFromShallowCopy().

Number of examples to use during each of the different greedy steps of the training phase.

Definition at line 122 of file GaussianDBNClassification.h.

Referenced by build_(), declareOptions(), makeDeepCopyFromShallowCopy(), and train().

Definition at line 132 of file GaussianDBNClassification.h.

Referenced by declareOptions(), greedyStep(), and jointGreedyStep().

The weight decay.

Definition at line 81 of file GaussianDBNClassification.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