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

#include <VariableSelectionWithDirectedGradientDescent.h>

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

List of all members.

Public Member Functions

 VariableSelectionWithDirectedGradientDescent ()
 Default constructor.
virtual string classname () const
virtual OptionListgetOptionList () const
virtual OptionMapgetOptionMap () const
virtual RemoteMethodMapgetRemoteMethodMap () const
virtual
VariableSelectionWithDirectedGradientDescent
deepCopy (CopiesMap &copies) const
virtual void makeDeepCopyFromShallowCopy (CopiesMap &copies)
 Transforms a shallow copy into a deep copy.
virtual void build ()
 Finish building the object; just call inherited::build followed by build_()
virtual void train ()
 *** SUBCLASS WRITING: ***
virtual void forget ()
 *** SUBCLASS WRITING: ***
virtual int outputsize () const
 SUBCLASS WRITING: override this so that it returns the size of this learner's output, as a function of its inputsize(), targetsize() and set options.
virtual TVec< string > getTrainCostNames () const
 *** SUBCLASS WRITING: ***
virtual TVec< string > getTestCostNames () const
 *** SUBCLASS WRITING: ***
virtual void computeOutput (const Vec &input, Vec &output) const
 *** SUBCLASS WRITING: ***
virtual void computeCostsFromOutputs (const Vec &input, const Vec &output, const Vec &target, Vec &costs) const
 *** SUBCLASS WRITING: ***
virtual void setTrainingSet (VMat training_set, bool call_forget=true)
 Overridden to take into account the new value of targetsize.

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 ()

Static Public Attributes

static StaticInitializer _static_initializer_

Static Protected Member Functions

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

Private Types

typedef PLearner inherited

Private Member Functions

void build_ ()
 **** SUBCLASS WRITING: ****
void verbose (string the_msg, int the_level)

Private Attributes

int inputsize
int targetsize
int weightsize
int length
int width
real learning_rate
Mat input_weights
TVec< boolweights_selected
TVec< intselected_variables
PP< ProgressBarpb
int row
int col
Vec sample_input
Vec sample_target
real sample_weight
Vec sample_output
Vec sample_cost
Vec train_criterion
real n7_value
real n8_value
real n9_value
real n10_value
real n10_gradient
real n9_gradient
real n8_gradient
real n7_gradient
Mat weights_gradient
int weights_gradient_max_col
real weights_gradient_max
Vec sum_of_abs_gradient
 Stores the sum of the absolute values of the gradient over all targets.

Detailed Description

Definition at line 51 of file VariableSelectionWithDirectedGradientDescent.h.


Member Typedef Documentation

Reimplemented from PLearn::PLearner.

Definition at line 53 of file VariableSelectionWithDirectedGradientDescent.h.


Constructor & Destructor Documentation

PLearn::VariableSelectionWithDirectedGradientDescent::VariableSelectionWithDirectedGradientDescent ( )

Default constructor.

Definition at line 48 of file VariableSelectionWithDirectedGradientDescent.cc.

    : learning_rate(1e-2)
{
}

Member Function Documentation

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

Reimplemented from PLearn::PLearner.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::PLearner.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::PLearner.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::PLearner.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::Object.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::PLearner.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Finish building the object; just call inherited::build followed by build_()

Reimplemented from PLearn::PLearner.

Definition at line 104 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Here is the call graph for this function:

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

**** SUBCLASS WRITING: ****

This method should finish building of the object, according to set 'options', in *any* situation.

Typical situations include:

  • Initial building of an object from a few user-specified options
  • Building of a "reloaded" object: i.e. from the complete set of all serialised options.
  • Updating or "re-building" of an object after a few "tuning" options (such as hyper-parameters) have been modified.

You can assume that the parent class' build_() has already been called.

A typical build method will want to know the inputsize(), targetsize() and outputsize(), and may also want to check whether train_set->hasWeights(). All these methods require a train_set to be set, so the first thing you may want to do, is check if(train_set), before doing any heavy building...

Note: build() is always called by setTrainingSet.

Reimplemented from PLearn::PLearner.

Definition at line 110 of file VariableSelectionWithDirectedGradientDescent.cc.

Referenced by build().

{
}

Here is the caller graph for this function:

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

Reimplemented from PLearn::Object.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

*** SUBCLASS WRITING: ***

This should be defined in subclasses to compute the weighted costs from already computed output. The costs should correspond to the cost names returned by getTestCostNames().

NOTE: In exotic cases, the cost may also depend on some info in the input, that's why the method also gets so see it.

Implements PLearn::PLearner.

Definition at line 304 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::exp(), PLearn::is_missing(), MISSING_VALUE, and pl_log.

Referenced by train().

{
    if (is_missing(outputv[0]))
    {
        costsv[0] = MISSING_VALUE;
        return;
        // ???  return MISSING_VALUE;
    }
    // Note that the "2 * target - 1" operation is only here to transform a 0/1
    // target into -1/1.
    costsv[0] = -pl_log(1.0 / (1.0 + exp(-(2.0 * targetv[0] - 1) * outputv[0])));;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::VariableSelectionWithDirectedGradientDescent::computeOutput ( const Vec input,
Vec output 
) const [virtual]

*** SUBCLASS WRITING: ***

This should be defined in subclasses to compute the output from the input.

Reimplemented from PLearn::PLearner.

Definition at line 289 of file VariableSelectionWithDirectedGradientDescent.cc.

References col, i, input_weights, PLearn::PLearner::inputsize(), PLearn::TVec< T >::resize(), and PLearn::PLearner::targetsize().

Referenced by train().

{
    outputv.resize(targetsize);
    for (int i = 0; i < targetsize; i++) {
        outputv[i] = input_weights(i, inputsize);
        for (int col = 0; col < inputsize; col++)
        {
            outputv[i] += input_weights(i, col) * inputv[col];
        }
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

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

Declares this class' options.

Reimplemented from PLearn::PLearner.

Definition at line 75 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::OptionBase::buildoption, PLearn::declareOption(), PLearn::PLearner::declareOptions(), input_weights, learning_rate, PLearn::OptionBase::learntoption, selected_variables, and weights_selected.

{ 
    declareOption(ol, "learning_rate", &VariableSelectionWithDirectedGradientDescent::learning_rate, OptionBase::buildoption,
                  "The learning rate of the gradient descent algorithm.\n");
    declareOption(ol, "input_weights", &VariableSelectionWithDirectedGradientDescent::input_weights, OptionBase::learntoption,
                  "The lerant weights of the linear probability estimator.\n");
    declareOption(ol, "weights_selected", &VariableSelectionWithDirectedGradientDescent::weights_selected, OptionBase::learntoption,
                  "The vector that identifies the non-zero weights.\n");
    declareOption(ol, "selected_variables", &VariableSelectionWithDirectedGradientDescent::selected_variables, OptionBase::learntoption,
                  "The vector with the selected variables in the order of their selection.\n");
    inherited::declareOptions(ol);
}

Here is the call graph for this function:

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

Reimplemented from PLearn::PLearner.

Definition at line 95 of file VariableSelectionWithDirectedGradientDescent.h.

:

VariableSelectionWithDirectedGradientDescent * PLearn::VariableSelectionWithDirectedGradientDescent::deepCopy ( CopiesMap copies) const [virtual]

Reimplemented from PLearn::PLearner.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

*** SUBCLASS WRITING: ***

(Re-)initializes the PLearner 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!)

A typical forget() method should do the following:

  • initialize the learner's parameters, using this random generator
  • stage = 0;

This method is typically called by the build_() method, after it has finished setting up the parameters, and if it deemed useful to set or reset the learner in its fresh state. (remember build may be called after modifying options that do not necessarily require the learner to restart from a fresh state...) forget is also called by the setTrainingSet method, after calling build(), so it will generally be called TWICE during setTrainingSet!

Reimplemented from PLearn::PLearner.

Definition at line 256 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::PLearner::inputsize(), PLearn::TVec< T >::resize(), selected_variables, and PLearn::PLearner::stage.

{
    inputsize = -1; // For safety reasons.
    selected_variables.resize(0);
    stage = 0;
}

Here is the call graph for this function:

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

Reimplemented from PLearn::Object.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::Object.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

Reimplemented from PLearn::Object.

Definition at line 73 of file VariableSelectionWithDirectedGradientDescent.cc.

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

*** SUBCLASS WRITING: ***

This should return the names of the costs computed by computeCostsFromOutputs.

Implements PLearn::PLearner.

Definition at line 281 of file VariableSelectionWithDirectedGradientDescent.cc.

References getTrainCostNames().

{ 
    return getTrainCostNames();
}

Here is the call graph for this function:

TVec< string > PLearn::VariableSelectionWithDirectedGradientDescent::getTrainCostNames ( ) const [virtual]

*** SUBCLASS WRITING: ***

This should return the names of the objective costs that the train method computes and for which it updates the VecStatsCollector train_stats.

Implements PLearn::PLearner.

Definition at line 271 of file VariableSelectionWithDirectedGradientDescent.cc.

Referenced by getTestCostNames().

{
    TVec<string> return_msg(1);
    return_msg[0] = "negloglikelihood";
    return return_msg;
}

Here is the caller graph for this function:

void PLearn::VariableSelectionWithDirectedGradientDescent::makeDeepCopyFromShallowCopy ( CopiesMap copies) [virtual]
int PLearn::VariableSelectionWithDirectedGradientDescent::outputsize ( ) const [virtual]

SUBCLASS WRITING: override this so that it returns the size of this learner's output, as a function of its inputsize(), targetsize() and set options.

Implements PLearn::PLearner.

Definition at line 263 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::PLearner::targetsize().

{
    return targetsize;
}

Here is the call graph for this function:

void PLearn::VariableSelectionWithDirectedGradientDescent::setTrainingSet ( VMat  training_set,
bool  call_forget = true 
) [virtual]

Overridden to take into account the new value of targetsize.

Reimplemented from PLearn::PLearner.

Definition at line 321 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::PLearner::setTrainingSet(), and PLearn::PLearner::targetsize().

                                                                                                     {
    targetsize = training_set->targetsize();
    inherited::setTrainingSet(training_set, call_forget);
}

Here is the call graph for this function:

void PLearn::VariableSelectionWithDirectedGradientDescent::train ( ) [virtual]

*** SUBCLASS WRITING: ***

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

TYPICAL CODE:

  static Vec input;  // static so we don't reallocate/deallocate 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;
  
  if(!train_stats)   // make a default stats collector, in case there's none
      train_stats = new VecStatsCollector();
  
  if(nstages<stage)  // asking to revert to a previous stage!
      forget();      // reset the learner to stage=0
  
  while(stage<nstages)
  {
      // clear statistics of previous epoch
      train_stats->forget(); 
            
      //... train for 1 stage, and update train_stats,
      // using train_set->getSample(input, target, weight);
      // and train_stats->update(train_costs)
          
      ++stage;
      train_stats->finalize(); // finalize statistics for this epoch
  }

Implements PLearn::PLearner.

Definition at line 117 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::TVec< T >::append(), col, computeCostsFromOutputs(), computeOutput(), PLearn::exp(), PLearn::fast_exact_is_equal(), PLearn::TVec< T >::fill(), PLearn::TMat< T >::fill(), PLearn::VMat::getExample(), i, input_weights, PLearn::PLearner::inputsize(), PLearn::is_missing(), j, learning_rate, PLearn::TVec< T >::length(), PLearn::VMat::length(), length, n10_gradient, n10_value, n7_gradient, n7_value, n8_gradient, n8_value, n9_gradient, n9_value, PLearn::PLearner::nstages, pb, pl_log, PLERROR, PLearn::PLearner::report_progress, PLearn::TVec< T >::resize(), PLearn::TMat< T >::resize(), row, sample_cost, sample_input, sample_output, sample_target, sample_weight, selected_variables, PLearn::PLearner::stage, sum_of_abs_gradient, PLearn::PLearner::targetsize(), PLearn::tostring(), train_criterion, PLearn::PLearner::train_set, PLearn::PLearner::train_stats, verbose(), weights_gradient, weights_gradient_max, weights_gradient_max_col, weights_selected, PLearn::PLearner::weightsize(), PLearn::PLearner::weightsize_, PLearn::VMat::width(), and width.

{
    if (!train_set)
        PLERROR("VariableSelectionWithDirectedGradientDescent: the algorithm has not been properly built");
    if (stage == 0) {
        // Initialize stuff before training.
        length = train_set->length();
        width = train_set->width();
        if (length < 1)
            PLERROR("VariableSelectionWithDirectedGradientDescent: the training set must contain at least one sample, got %d", length);
        inputsize = train_set->inputsize();
        targetsize = train_set->targetsize();
        weightsize = train_set->weightsize();
        if (inputsize < 1)
            PLERROR("VariableSelectionWithDirectedGradientDescent: expected  inputsize greater than 0, got %d", inputsize);
        if (targetsize <= 0)
            PLERROR("In VariableSelectionWithDirectedGradientDescent::train - The targetsize (%d) must be >= 1", targetsize);
        if (weightsize != 0)
            PLERROR("VariableSelectionWithDirectedGradientDescent: expected weightsize to be 1, got %d", weightsize_);
        input_weights.resize(targetsize, inputsize + 1);
        weights_selected.resize(inputsize + 1);
        weights_gradient.resize(targetsize, inputsize + 1);
        sample_input.resize(inputsize);
        sample_target.resize(1);
        sample_output.resize(1);
        sample_cost.resize(1);
        train_criterion.resize(targetsize);
        sum_of_abs_gradient.resize(inputsize);
    }

    input_weights.fill(0);
    weights_selected.fill(false);
    if (report_progress)
    {
        pb = new ProgressBar("VariableSelectionWithDirectedGradientDescent : train stages: ", nstages);
    }
/*
  We loop through the data for the specified maximum number of stages.
*/
    for (; stage < nstages; stage++)
    {
        weights_gradient.fill(0);
/*
  We compute the train criterion for this stage and compute the weight gradient.
*/
        train_criterion.fill(0);
        for (int i = 0; i < targetsize; i++) {
            for (row = 0; row < length; row++)
            {
                real target = train_set(row, inputsize + i);
                if (is_missing(target))
                    continue;
                n7_value = input_weights(i, inputsize);
                for (col = 0; col < inputsize; col++)
                {
                    n7_value += input_weights(i, col) * train_set(row, col);
                }
#ifdef BOUNDCHECK
                if (!fast_exact_is_equal(target, 0.0) &&
                    !fast_exact_is_equal(target, 1.0))
                    PLERROR("In VariableSelectionWithDirectedGradientDescent::train - The target should be 0 or 1");
#endif
                if (fast_exact_is_equal(target, 0)) target = -1; // We work with -1 and 1 instead.
                n8_value = target * n7_value;
                n9_value = 1.0 / (1.0 + exp(-n8_value));
                n10_value = -pl_log(n9_value);
                train_criterion[i] += n10_value;
                n10_gradient = 1.0;
                n9_gradient = n10_gradient * (-1.0 / n9_value);
                n8_gradient = n9_gradient * n9_value * 1.0 / (1.0 + exp(n8_value));
                n7_gradient = n8_gradient * target;
                for (col = 0; col < inputsize; col++)
                {
                    weights_gradient(i, col) += n7_gradient * train_set(row, col);
                }
                weights_gradient(i, inputsize) += n7_gradient;     
            }
        }
/*
  We perform this stage weight update according to the directed gradient descent algorithm.
*/
        sum_of_abs_gradient.fill(0);
        for (int i = 0; i < targetsize; i++) {
            // Bias update.
            input_weights(i, inputsize) -= learning_rate * weights_gradient(i, inputsize);
            // Compute sum of |gradient|.
            for (int j = 0; j < inputsize; j++)
                sum_of_abs_gradient[j] += fabs(weights_gradient(i,j));
        }
        weights_gradient_max = 0.0;
        for (col = 0; col < inputsize; col++)
        {
            if (sum_of_abs_gradient[col] > weights_gradient_max)
            {
                weights_gradient_max = sum_of_abs_gradient[col];
                weights_gradient_max_col = col;
            }
        }
        if (!weights_selected[weights_gradient_max_col])
        {
            selected_variables.append(weights_gradient_max_col);
            verbose("VariableSelectionWithDirectedGradientDescent: variable " + tostring(weights_gradient_max_col)
                    + " was added.", 2);
        }
        weights_selected[weights_gradient_max_col] = true;
        // Weights update.
        for (int i = 0; i < targetsize; i++)
            for (col = 1; col < inputsize; col++)
                input_weights(i, col) -= learning_rate * weights_gradient(i, col) * real(weights_selected[col]);
        verbose("VariableSelectionWithDirectedGradientDescent: After " + tostring(stage) + " stages, the train criterion is: "
                + tostring(train_criterion), 3);
        if (report_progress) pb->update(stage);
    }
    if (report_progress)
    {
        pb = new ProgressBar("VariableSelectionWithDirectedGradientDescent : computing the training statistics: ", length);
    }
    train_stats->forget();
    for (row = 0; row < length; row++)
    {   
        train_set->getExample(row, sample_input, sample_target, sample_weight);
        for (int i = 0; i < sample_target.length(); i++)
            if (fast_exact_is_equal(sample_target[i], 0)) sample_target[i] = -1; // We work with -1 and 1.
        computeOutput(sample_input, sample_output);
        computeCostsFromOutputs(sample_input, sample_output, sample_target, sample_cost);
        train_stats->update(sample_cost);
        if (report_progress) pb->update(row);
    }
    train_stats->finalize();
    verbose("VariableSelectionWithDirectedGradientDescent: After " + tostring(stage) + " stages, average error is: "
            + tostring(train_stats->getMean()), 1);
}

Here is the call graph for this function:

void PLearn::VariableSelectionWithDirectedGradientDescent::verbose ( string  the_msg,
int  the_level 
) [private]

Definition at line 250 of file VariableSelectionWithDirectedGradientDescent.cc.

References PLearn::endl(), PLearn::pout, and PLearn::PLearner::verbosity.

Referenced by train().

{
    if (verbosity >= the_level)
        pout << the_msg << endl;
}

Here is the call graph for this function:

Here is the caller graph for this function:


Member Data Documentation

Reimplemented from PLearn::PLearner.

Definition at line 95 of file VariableSelectionWithDirectedGradientDescent.h.

Definition at line 68 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by computeOutput(), and train().

Definition at line 61 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by declareOptions(), and train().

Definition at line 58 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 79 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 78 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 82 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 75 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 81 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 76 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 80 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 77 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 66 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 67 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Stores the sum of the absolute values of the gradient over all targets.

Definition at line 87 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by makeDeepCopyFromShallowCopy(), and train().

Definition at line 85 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 84 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().

Definition at line 59 of file VariableSelectionWithDirectedGradientDescent.h.

Referenced by train().


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