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

#include <CCCostVariable.h>

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

List of all members.

Public Member Functions

 CCCostVariable ()
 protected default constructor for persistence
 CCCostVariable (VMat the_distr, Func the_f, Func f_candidate)
 constructor
virtual string classname () const
virtual OptionListgetOptionList () const
virtual OptionMapgetOptionMap () const
virtual RemoteMethodMapgetRemoteMethodMap () const
virtual CCCostVariabledeepCopy (CopiesMap &copies) const
virtual void build ()
 Post-constructor.
virtual void recomputeSize (int &l, int &w) const
 Recomputes the length l and width w that this variable should have, according to its parent variables.
virtual void makeDeepCopyFromShallowCopy (CopiesMap &copies)
 Does the necessary operations to transform a shallow copy (this) into a deep copy by deep-copying all the members that need to be.
virtual void fprop ()
 compute output given input
virtual void bprop ()
virtual void fbprop ()
 do both fprop and bprop
virtual void symbolicBprop ()
 compute a piece of new Var graph that represents the symbolic derivative of this Var
virtual void rfprop ()
void printInfo (bool print_gradient)

Static Public Member Functions

static string _classname_ ()
 CCCostVariable.
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 void declareOptions (OptionList &ol)
 Default constructor for persistence.

Public Attributes

VMat distr
Func f_error
Func f_candidate
Vec input_value
Vec input_gradient
Vec error_output_value
Vec candidate_output_value
Vec error_correlations
Vec adjusted_gradient

Static Public Attributes

static StaticInitializer _static_initializer_

Protected Member Functions

void build_ ()
 Object-specific post-constructor.

Private Types

typedef NaryVariable inherited

Private Attributes

Vec mean_error
real mean_candidate

Detailed Description

Definition at line 52 of file CCCostVariable.h.


Member Typedef Documentation

Reimplemented from PLearn::NaryVariable.

Definition at line 54 of file CCCostVariable.h.


Constructor & Destructor Documentation

PLearn::CCCostVariable::CCCostVariable ( ) [inline]

protected default constructor for persistence

Definition at line 74 of file CCCostVariable.h.

: distr(), f_error(), f_candidate() {}
PLearn::CCCostVariable::CCCostVariable ( VMat  the_distr,
Func  the_f,
Func  f_candidate 
)

constructor

Definition at line 58 of file CCCostVariable.cc.

References build_().

    : inherited(nonInputParentsOfPath(the_f_candidate->inputs,the_f_candidate->outputs), 
                1,
                1),
      distr(the_distr), f_error(the_f_error), f_candidate(the_f_candidate),
      input_value(the_distr->width()),
      input_gradient(distr->inputsize()),
      error_output_value(the_f_error->outputs[0]->size()),
      candidate_output_value(the_f_candidate->outputs[0]->size()),
      error_correlations(the_f_error->outputs[0]->size()),
      adjusted_gradient(1)
{
    build_();
}

Here is the call graph for this function:


Member Function Documentation

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

CCCostVariable.

Reimplemented from PLearn::NaryVariable.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::NaryVariable.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::NaryVariable.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::NaryVariable.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::Object.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::NaryVariable.

Definition at line 56 of file CCCostVariable.cc.

void PLearn::CCCostVariable::bprop ( ) [virtual]

Implements PLearn::Variable.

Definition at line 169 of file CCCostVariable.cc.

References fbprop().

{ fbprop(); }

Here is the call graph for this function:

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

Post-constructor.

The normal implementation should call simply inherited::build(), then this class's build_(). This method should be callable again at later times, after modifying some option fields to change the "architecture" of the object.

Reimplemented from PLearn::NaryVariable.

Definition at line 74 of file CCCostVariable.cc.

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

Here is the call graph for this function:

void PLearn::CCCostVariable::build_ ( ) [protected]

Object-specific post-constructor.

This method should be redefined in subclasses and do the actual building of the object according to previously set option fields. Constructors can just set option fields, and then call build_. This method is NOT virtual, and will typically be called only from three places: a constructor, the public virtual build() method, and possibly the public virtual read method (which calls its parent's read). build_() can assume that its parent's build_() has already been called.

Reimplemented from PLearn::NaryVariable.

Definition at line 81 of file CCCostVariable.cc.

References distr, f_candidate, f_error, input_gradient, input_value, mean_error, PLERROR, and PLearn::TVec< T >::resize().

Referenced by build(), and CCCostVariable().

{
    if (f_error && f_candidate && distr) {
        mean_error.resize(f_error->outputs[0]->size());
        input_value.resize(distr->inputsize() + distr->targetsize() + distr->weightsize());
        input_gradient.resize(distr->inputsize());
        if(f_error->outputs.size() != 1)
            PLERROR("In CCCostVariable: error function must have a single variable output (maybe you can vconcat the vars into a single one prior to calling sumOf, if this is really what you want)");
      
        if(f_error->outputs[0].width() != 1)
            PLERROR("In CCCostVariable: the error function's output must be a column vector ");
        f_error->inputs.setDontBpropHere(true);
      
        if(f_candidate->outputs.size() != 1)
            PLERROR("In CCCostVariable: candidate node function must have a single variable output (maybe you can vconcat the vars into a single one prior to calling sumOf, if this is really what you want)");
      
        if(f_candidate->outputs[0].width() != 1 || f_candidate->outputs[0].length() != 1)
            PLERROR("In CCCostVariable: the candidate node function's output must be a column vector ");
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

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

Reimplemented from PLearn::Object.

Definition at line 56 of file CCCostVariable.cc.

void PLearn::CCCostVariable::declareOptions ( OptionList ol) [static]

Default constructor for persistence.

Reimplemented from PLearn::NaryVariable.

Definition at line 103 of file CCCostVariable.cc.

References PLearn::OptionBase::buildoption, PLearn::declareOption(), PLearn::NaryVariable::declareOptions(), distr, f_candidate, and f_error.

Here is the call graph for this function:

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

Reimplemented from PLearn::NaryVariable.

Definition at line 78 of file CCCostVariable.h.

:
    void build_();
CCCostVariable * PLearn::CCCostVariable::deepCopy ( CopiesMap copies) const [virtual]

Reimplemented from PLearn::NaryVariable.

Definition at line 56 of file CCCostVariable.cc.

void PLearn::CCCostVariable::fbprop ( ) [virtual]

do both fprop and bprop

Reimplemented from PLearn::Variable.

Definition at line 173 of file CCCostVariable.cc.

References adjusted_gradient, candidate_output_value, distr, error_correlations, error_output_value, f_candidate, f_error, fprop(), PLearn::Variable::gradient, i, input_gradient, input_value, j, PLearn::TVec< T >::length(), PLearn::VMat::length(), mean_error, PLearn::TVec< T >::resize(), PLearn::TVec< T >::subVec(), and PLearn::VMat::width().

Referenced by bprop().

{
    fprop();

    for(int i=0; i<distr->length(); i++)
    {
        input_value.resize(distr->width());
        distr->getRow(i, input_value);
        input_value.resize(distr->inputsize()+distr->targetsize()+distr->weightsize());
        f_error->fprop(input_value, error_output_value);
        for(int j=0; j<error_correlations.length(); j++)
        {
      
            adjusted_gradient[0] = -1*gradient[0]*(error_output_value[j]-mean_error[j])
                * (error_correlations[j] > 0 ? 1 : -1)/distr->length();
            f_candidate->fbprop(input_value.subVec(0,distr->inputsize()),candidate_output_value,input_gradient, adjusted_gradient); // could be more efficient (do just a bprop: not implemented) 
        }
    }

}

Here is the call graph for this function:

Here is the caller graph for this function:

void PLearn::CCCostVariable::fprop ( ) [virtual]

compute output given input

Implements PLearn::Variable.

Definition at line 129 of file CCCostVariable.cc.

References PLearn::abs(), candidate_output_value, PLearn::TVec< T >::clear(), distr, error_correlations, error_output_value, f_candidate, f_error, i, input_value, j, PLearn::TVec< T >::length(), PLearn::VMat::length(), mean_candidate, mean_error, PLearn::TVec< T >::resize(), PLearn::TVec< T >::subVec(), PLearn::Variable::value, and PLearn::VMat::width().

Referenced by fbprop(), and printInfo().

{
    f_error->recomputeParents();
    f_candidate->recomputeParents();
    mean_error.clear();
    mean_candidate=0;

    // Compute the means of the candidate node and of the error for every output
    for(int i=0; i<distr->length(); i++)
    {
        input_value.resize(distr->width());
        distr->getRow(i, input_value);
        input_value.resize(distr->inputsize()+distr->targetsize()+distr->weightsize());
        f_error->fprop(input_value, error_output_value);
        mean_error += error_output_value;
        f_candidate->fprop(input_value.subVec(0,distr->inputsize()),candidate_output_value);
        mean_candidate += candidate_output_value[0];
    }

    mean_error /= distr->length();
    mean_candidate /= distr->length();

    value.clear();
    error_correlations.clear();
    for(int i=0; i<distr->length(); i++)
    {
        input_value.resize(distr->width());
        distr->getRow(i, input_value);
        input_value.resize(distr->inputsize()+distr->targetsize()+distr->weightsize());
        f_error->fprop(input_value, error_output_value);
        f_candidate->fprop(input_value.subVec(0,distr->inputsize()),candidate_output_value);
        for(int j=0; j<error_correlations.length(); j++)
            error_correlations[j] += (candidate_output_value[0]-mean_candidate)*(error_output_value[j]-mean_error[j]);
    }
    for(int j=0; j<error_correlations.length(); j++)
        value[0] -= abs(error_correlations[j]);
    value[0] /= distr->length();
}

Here is the call graph for this function:

Here is the caller graph for this function:

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

Reimplemented from PLearn::Object.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::Object.

Definition at line 56 of file CCCostVariable.cc.

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

Reimplemented from PLearn::Object.

Definition at line 56 of file CCCostVariable.cc.

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

Does the necessary operations to transform a shallow copy (this) into a deep copy by deep-copying all the members that need to be.

This needs to be overridden by every class that adds "complex" data members to the class, such as Vec, Mat, PP<Something>, etc. Typical implementation:

  void CLASS_OF_THIS::makeDeepCopyFromShallowCopy(CopiesMap& copies)
  {
      inherited::makeDeepCopyFromShallowCopy(copies);
      deepCopyField(complex_data_member1, copies);
      deepCopyField(complex_data_member2, copies);
      ...
  }
Parameters:
copiesA map used by the deep-copy mechanism to keep track of already-copied objects.

Reimplemented from PLearn::NaryVariable.

Definition at line 120 of file CCCostVariable.cc.

References PLearn::deepCopyField(), distr, f_candidate, f_error, and PLearn::NaryVariable::makeDeepCopyFromShallowCopy().

Here is the call graph for this function:

void PLearn::CCCostVariable::printInfo ( bool  print_gradient) [virtual]

Reimplemented from PLearn::NaryVariable.

Definition at line 206 of file CCCostVariable.cc.

References adjusted_gradient, candidate_output_value, distr, PLearn::endl(), error_correlations, error_output_value, f_candidate, f_error, fprop(), PLearn::Variable::getName(), PLearn::Variable::gradient, i, PLearn::Object::info(), input_gradient, input_value, j, PLearn::TVec< T >::length(), PLearn::VMat::length(), mean_error, PLearn::TVec< T >::resize(), PLearn::TVec< T >::subVec(), PLearn::Variable::value, and PLearn::VMat::width().

{
  
    fprop();

    for(int i=0; i<distr->length(); i++)
    {
        input_value.resize(distr->width());
        distr->getRow(i, input_value);
        input_value.resize(distr->inputsize()+distr->targetsize()+distr->weightsize());
        f_error->fprop(input_value, error_output_value);
        for(int j=0; j<error_correlations.length(); j++)
        {
      
            adjusted_gradient[0] = -1*gradient[0]*(error_output_value[j]-mean_error[j])
                * (error_correlations[j] > 0 ? 1 : -1);
            f_candidate->fbprop(input_value.subVec(0,distr->inputsize()),candidate_output_value,input_gradient, adjusted_gradient); // could be more efficient 
        }
        f_candidate->fproppath.printInfo(print_gradient);
    }
  
    cout << info() << " : " << getName() << " = " << value;
    if (print_gradient) cout << " gradient=" << gradient;
    cout << endl; 
}

Here is the call graph for this function:

void PLearn::CCCostVariable::recomputeSize ( int l,
int w 
) const [virtual]

Recomputes the length l and width w that this variable should have, according to its parent variables.

This is used for ex. by sizeprop() The default version stupidly returns the current dimensions, so make sure to overload it in subclasses if this is not appropriate.

Reimplemented from PLearn::Variable.

Definition at line 113 of file CCCostVariable.cc.

{
    w = 1;
    l = 1;
}
void PLearn::CCCostVariable::rfprop ( ) [virtual]

Reimplemented from PLearn::Variable.

Definition at line 200 of file CCCostVariable.cc.

References PLERROR.

{
    PLERROR("In CCCostVariable::rfprop() : Not implemented");
}
void PLearn::CCCostVariable::symbolicBprop ( ) [virtual]

compute a piece of new Var graph that represents the symbolic derivative of this Var

Reimplemented from PLearn::Variable.

Definition at line 195 of file CCCostVariable.cc.

References PLERROR.

{
    PLERROR("In CCCostVariable::symbolicBprop() : Not implemented");
}

Member Data Documentation

Reimplemented from PLearn::NaryVariable.

Definition at line 78 of file CCCostVariable.h.

Definition at line 67 of file CCCostVariable.h.

Referenced by fbprop(), and printInfo().

Definition at line 65 of file CCCostVariable.h.

Referenced by fbprop(), fprop(), and printInfo().

Definition at line 66 of file CCCostVariable.h.

Referenced by fbprop(), fprop(), and printInfo().

Definition at line 64 of file CCCostVariable.h.

Referenced by fbprop(), fprop(), and printInfo().

Definition at line 63 of file CCCostVariable.h.

Referenced by build_(), fbprop(), and printInfo().

Definition at line 62 of file CCCostVariable.h.

Referenced by build_(), fbprop(), fprop(), and printInfo().

Definition at line 70 of file CCCostVariable.h.

Referenced by fprop().

Definition at line 69 of file CCCostVariable.h.

Referenced by build_(), fbprop(), fprop(), and printInfo().


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