#include <code_book.h>

Inheritance diagram for CodeBook:

Collaboration diagram for CodeBook:

Public Member Functions
	CodeBook (const bool &_calib=false)

	CodeBook (unsigned _n, unsigned _size, bool _cal=false)

	CodeBook (unsigned _n, unsigned _size, floatFeature _lower=0, floatFeature _upper=1, bool _cal=false)

	CodeBook (unsigned _n, const ClassicTrainingVectors &_ts, const bool _use_rand_cvs)

	CodeBook (std::istream &_is)

virtual	~CodeBook ()

virtual FeatureVector &	test (const FeatureVector &_in) const

virtual unsigned	testIndex (const FeatureVector &_in) const

virtual void	classify (const ClassicTrainingVectors *_ts)

virtual const std::vector< unsigned > &	classifAt (const unsigned &_index) const

virtual unsigned	classifSizeAt (const unsigned &_index) const

virtual Label	apply (const FeatureVector &_in) const

virtual void	calibrate (ClassicTrainingVectors &_ts, Label _def="")

virtual unsigned	output (const FeatureVector &_in) const

virtual void	printSelf (std::ostream &_os) const

virtual void	readSelf (std::istream &_is, long _dim=-1, long _size=-1)

virtual void	saveObject (std::ostream &_os) const

virtual void	loadObject (std::istream &_is)

virtual void	Normalize (const std::vector< ClassicTrainingVectors::statsStruct > &_varStats)

virtual void	unNormalize (const std::vector< ClassicTrainingVectors::statsStruct > &_varStats)

virtual void	printHistogram (std::ostream &_os) const

virtual void	printQuantError (std::ostream &_os) const

Public Member Functions inherited from ClassificationDataSet< FeatureVector, Label >
	ClassificationDataSet ()

virtual	~ClassificationDataSet ()

Public Member Functions inherited from ClassificationTrainingSet< FeatureVector, Label >
	ClassificationTrainingSet (const bool &_calib=true, unsigned _n=0)

	ClassificationTrainingSet (std::istream &_is)

virtual	~ClassificationTrainingSet ()

void	setSplit (float _tp, float _vp)

splitIt	beginSubset (unsigned _um)

splitIt	endSubset (unsigned _um)
	Returns an iterator to the end of the subset. More...

virtual void	add (const FeatureVector &_i, const Label &_tg)

virtual void	add (const FeatureVector &_i)

virtual bool	remove (unsigned int _idx)

size_t	size () const

const Label &	targetAt (unsigned _i) const

Label &	targetAt (unsigned _i)

const FeatureVector &	itemAt (unsigned _i) const

FeatureVector &	itemAt (unsigned _i)

bool	calibrated () const

void	calibrated (const bool &_calib)

void	clear ()

virtual void	readSelf (std::istream &_is)

unsigned	numTargets () const

virtual bool	swapItems (unsigned _i, unsigned _j)

Public Attributes
std::vector< std::vector< unsigned > >	classifVectors

std::vector< double >	aveDistances

Public Attributes inherited from ClassificationTrainingSet< FeatureVector, Label >
std::vector< FeatureVector >	theItems

std::vector< Label >	theTargets

Protected Member Functions
void	readClassifVectors (std::istream &_is)

void	writeClassifVectors (std::ostream &_os) const

Protected Member Functions inherited from ClassificationTrainingSet< FeatureVector, Label >
void	computeNumTargets ()

void	checkCalibrated (std::istream &_is)

void	readItems (std::istream &_is)

void	writeCalibrated (std::ostream &_os) const

void	writeItems (std::ostream &_os, bool _delim=false) const

void	skipComments (std::istream &_is) const

std::vector< FeatureVector >::const_iterator	itemsBegin () const

std::vector< FeatureVector >::const_iterator	itemsEnd () const

std::vector< Label >::const_iterator	targetsBegin () const

std::vector< Label >::const_iterator	targetsEnd () const

Additional Inherited Members
Public Types inherited from ClassificationDataSet< FeatureVector, Label >
typedef FeatureVector	In
	Class of input vectors. Usually a FeatureVector (vector of Feature) More...

typedef Label	Out
	Class of the target. Can be a double, string, unsigned, even a vector ... More...

typedef ClassificationTrainingSet< In, Out >	TS
	Training set. Set of vectors (training vectors), probably classified. More...

Public Types inherited from ClassificationTrainingSet< FeatureVector, Label >
enum	splitMode
	Ways the training set can be used. More...

enum	useMode
	use of samples More...

typedef std::multimap< unsigned, unsigned, std::less< unsigned > >	splitTS
	Training sets mode. More...

typedef splitTS::iterator	splitIt
	iterator More...

Protected Attributes inherited from ClassificationTrainingSet< FeatureVector, Label >
bool	isCalibrated

splitTS	splitTrainingSet

unsigned	nTargets

Detailed Description

This class implements a codebook. A codebook is a set of examples (each of them being a vector). These examples are usually labeled, ie, classified (the codebook is calibrated), but it is not necessary (the codebook is NOT calibrated). A codebook is the result obtained after one has trained some kind of algorithms. The way to classify a data once the algorithm has been trained is to look for the example in the code book that best matches the data to classify. Then, the same label of the example in the codebook is associated to the data wanted to be classified (if it is a calibrated codebook), or the example itself is returned (if it is a NO calibrated codebook) indicating with this that the data belongs to the same 'class' that the returned example.

Definition at line 57 of file code_book.h.

Constructor & Destructor Documentation

◆ CodeBook() [1/5]

CodeBook::CodeBook ( const bool & _calib = false )

inline

Default constructor Parameter: _calib Calibrated or not, that is, a CB with class labels or not

Definition at line 69 of file code_book.h.

                                          : ClassificationDataSet<FeatureVector, Label>(),
             ClassificationTrainingSet<FeatureVector, Label>(_calib)
     {};

◆ CodeBook() [2/5]

CodeBook::CodeBook	(	unsigned	_n,
		unsigned	_size,
		bool	_cal = `false`
	)

Constructor. Constructs a codebook with initial code vectors at zero. from an unsigned integer to instantiate the template Parameter: _n Number of vectors Parameter: _size Size of code vectors Parameter: _cal Calibrated or not, that is, a CB with class labels or not

This class implements a codebook. A codebook is a set of examples (each of them being a vector). These examples are usually labeled, ie, classified (the codebook is calibrated), but it is not necessary (the codebook is NOT calibrated). A codebook is the result obtained after one has trained some kind of algorithms. The way to classify a data once the algorithm has been trained is to look for the example in the code book that best matches the data to classify. Then, the same label of the example in the codebook is associated to the data wanted to be classified (if it is a calibrated codebook), or the example itself is returned (if it is a NO calibrated codebook) indicating with this that the data belongs to the same 'class' that the returned example. Constructor. Constructs a codebook with initial code vectors at zero. from an unsigned integer to instantiate the template Parameter: _n Number of vectors Parameter: _size Size of code vectors Parameter: _lower Lower value for random elements Parameter: _upper Upper value for random elements Parameter: _cal Calibrated or not, that is, a CB with class labels or not

Definition at line 59 of file code_book.cpp.

         : ClassificationDataSet<FeatureVector, Label>(),
         ClassificationTrainingSet<FeatureVector, Label>(_cal)
 {
     // Fill vectors with zeros
     theItems.resize(_n);
     //if (calibrated())
     theTargets.resize(_n, "");
     for (unsigned i = 0 ; i < _n ; i++)
     {
         FeatureVector v;
         v.resize(_size, 0);
         theItems[i] = v;
     }
 }

◆ CodeBook() [3/5]

CodeBook::CodeBook	(	unsigned	_n,
		unsigned	_size,
		floatFeature	_lower = `0`,
		floatFeature	_upper = `1`,
		bool	_cal = `false`
	)

Constructor. Constructs a codebook with random initial code vectors. from an unsigned integer to instantiate the template Parameter: _n Number of vectors Parameter: _size Size of code vectors Parameter: _lower Lower value for random elements Parameter: _upper Upper value for random elements Parameter: _cal Calibrated or not, that is, a CB with class labels or not

Definition at line 87 of file code_book.cpp.

         : ClassificationDataSet<FeatureVector, Label>(),
         ClassificationTrainingSet<FeatureVector, Label>(_cal)
 {
     theItems.resize(_n);
     //if (calibrated())
     theTargets.resize(_n, "");
     // Assign random vectors
     for (unsigned i = 0 ; i < _n ; i++)
     {
         std::vector<floatFeature> v;
         v = randomVector(_size, _lower, _upper);
         theItems[i] = v;
     }
 }

◆ CodeBook() [4/5]

CodeBook::CodeBook	(	unsigned	_n,
		const ClassicTrainingVectors &	_ts,
		const bool	_use_rand_cvs
	)

Constructor. Constructs a codebook with initial code vectors taken randomly from the training file. from an unsigned integer to instantiate the template Parameter: _n Number of vectors Parameter: _ts Training set; will be used to get initial values Parameter: _use_rand_cvs Use random code vector values

Definition at line 119 of file code_book.cpp.

         : ClassificationDataSet<FeatureVector, Label>(),
         ClassificationTrainingSet<FeatureVector, Label>(_ts.calibrated())
 {
     // Take random samples
     //    RandomUniformGenerator<unsigned> chosen( 0, _ts.size() -1 );
     randomize_random_generator();
 
     theItems.resize(_n);
     //if (_ts.calibrated())
     theTargets.resize(_n, "");
     // Assign random vectors
     std::vector<floatFeature> v;
     for (unsigned i = 0 ; i < _n ; i++)
     {
         auto index = (int) rnd_unif(0, _ts.size() - 1);
         v = _ts.theItems[index];
         if (_use_rand_cvs)
         {
             // NT: Scan this vector for the range of pixel values
             floatFeature minval;
             floatFeature maxval;
             std::vector<floatFeature>::const_iterator viter = v.begin();
             minval = maxval = *viter;
             for (viter++; viter != v.end(); viter++)
             {
                 if (*viter < minval)
                     minval = *viter;
                 else if (*viter > maxval)
                     maxval = *viter;
             }
             v = randomVector(_ts.theItems[0].size(), minval, maxval);
         }
         theItems[i] = v;
     }
 
 }

◆ CodeBook() [5/5]

CodeBook::CodeBook ( std::istream & _is )

Constructs a code book given a stream Parameter: _is The input stream

Exceptions

runtime_error If there are problems with the stream

Definition at line 162 of file code_book.cpp.

                                   : ClassificationDataSet<FeatureVector, Label>(),
         ClassificationTrainingSet<FeatureVector, Label>(_is)
 {
     readSelf(_is);
 }

◆ ~CodeBook()

virtual CodeBook::~CodeBook ( )

inlinevirtual

Virtual destructor needed

Definition at line 117 of file code_book.h.

118 {};

Member Function Documentation

◆ apply()

Label CodeBook::apply ( const FeatureVector & _in ) const

virtual

Returns the label associated to an input Parameter: _in Sample to classify

Implements ClassificationDataSet< FeatureVector, Label >.

Definition at line 316 of file code_book.cpp.

 {
     return theTargets[testIndex(_in)];
 }

◆ calibrate()

void CodeBook::calibrate	(	ClassicTrainingVectors &	_ts,
		Label	_def = `""`
	)

virtual

Calibrates the code book Parameter: _ts The calibrated training set Parameter: _def Default target for non-calibrated vectors

Exceptions

runtime_error If the training set is not calibrated

Definition at line 328 of file code_book.cpp.

 {
     // set the default label
     for (std::vector<FeatureVector>::const_iterator i = itemsBegin() ;
          i < itemsEnd() ; i++)
         theTargets[i - itemsBegin()] = _def;
     if (_ts.calibrated())
     {
         for (unsigned j = 0 ; j < _ts.size() ; j++)
             theTargets[testIndex(_ts.theItems[j])] = _ts.theTargets[j];
         calibrated(true);
     }
     else
         calibrated(false);
 }

◆ classifAt()

const std::vector< unsigned > & CodeBook::classifAt ( const unsigned & _index ) const

virtual

Returns the list of input vectors associated to this code vector.

Definition at line 286 of file code_book.cpp.

 {
     if (_index < 0 || _index > classifVectors.size())
     {
         std::ostringstream msg;
         msg << "index out of range";
         throw std::runtime_error(msg.str());
     }
     return classifVectors[_index];
 }

◆ classifSizeAt()

unsigned CodeBook::classifSizeAt ( const unsigned & _index ) const

virtual

Returns the number of input vectors associated to this code vector.

Definition at line 300 of file code_book.cpp.

 {
     if (_index < 0 || _index > classifVectors.size())
     {
         std::ostringstream msg;
         msg << "index out of range";
         throw std::runtime_error(msg.str());
     }
     return classifVectors[_index].size();
 }

◆ classify()

void CodeBook::classify ( const ClassicTrainingVectors * _ts )

virtual

Fills the classifVectors with the list of the best input vectors associated to it. Parameter: _ts Sample list to classify

Reimplemented in FuzzyCodeBook.

Definition at line 238 of file code_book.cpp.

 {
     classifVectors.clear(); // clear previous classification.
     classifVectors.resize(size());
     aveDistances.clear(); // clear previous classification.
     aveDistances.resize(size());
     for (unsigned j = 0 ; j < _ts->size() ; j++)
         classifVectors[testIndex(_ts->theItems[j])].push_back(j);
 
     for (unsigned i = 0 ; i < size() ; i++)
     {
         double aveDist = 0;
         for (unsigned j = 0 ; j < classifVectors[i].size() ; j++)
             aveDist += euclideanDistance(theItems[i], _ts->theItems[classifVectors[i][j]]);
         if (classifVectors[i].size() != 0)
             aveDist /= (double) classifVectors[i].size();
         aveDistances[i] = (double) aveDist;
     }
 
 }

◆ loadObject()

void CodeBook::loadObject ( std::istream & _is )

virtual

Loads the xmippCodeBook class from a stream. this method can be used to load the status of the class. Parameter: _is The output stream

Loads the CodeBook class from a stream. this method can be used to load the status of the class. Parameter: _is The output stream

Reimplemented from ClassificationTrainingSet< FeatureVector, Label >.

Reimplemented in FuzzyMap, ClassificationMap, and FuzzyCodeBook.

Definition at line 478 of file code_book.cpp.

 {
     clear();
     readClassifVectors(_is);
     ClassificationTrainingSet<FeatureVector, Label>::loadObject(_is);
 }

◆ Normalize()

void CodeBook::Normalize ( const std::vector< ClassicTrainingVectors::statsStruct > & _varStats )

virtual

Normalize all features in the codebook Parameter: _varStats The normalization information

Definition at line 516 of file code_book.cpp.

 {
     using namespace std;
     if (_varStats.size() != theItems[0].size())
     {
         std::ostringstream msg;
         msg << "Normalization information does not coincide with codebook structure";
         throw std::runtime_error(msg.str());
     }
     for (unsigned it = 0; it < size(); it++)
     {
         for (unsigned i = 0; i < theItems[0].size(); i++)
         {
             if (!isnan(theItems[it][i]))
             {
                 if (_varStats[i].sd != 0)
                     theItems[it][i] = (theItems[it][i] - _varStats[i].mean) / _varStats[i].sd;
             }
         }
     }
 }

◆ output()

unsigned CodeBook::output ( const FeatureVector & _in ) const

virtual

Returns the index of the codevector closest to an input. This is the method used to classify inputs Parameter: _in Sample to classify.

Implements ClassificationDataSet< FeatureVector, Label >.

Definition at line 350 of file code_book.cpp.

 {
     return testIndex(_in);
 }

◆ printHistogram()

void CodeBook::printHistogram ( std::ostream & _os ) const

virtual

Prints the histogram values of each codevector. Parameter: _os The the output stream

Prints the histogram values of each Fuzzy codevector. Parameter: _os The the output stream

Definition at line 264 of file code_book.cpp.

 {
     _os << "1 " << size() << std::endl;
     for (size_t j = 0; j < size(); j++)
         _os << j << " " << classifSizeAt(j) << std::endl;
 }

◆ printQuantError()

void CodeBook::printQuantError ( std::ostream & _os ) const

virtual

Prints the Average Quantization Error of each codevector. Parameter: _os The the output stream

Definition at line 276 of file code_book.cpp.

 {
     _os << "1 " << size() << std::endl;
     for (size_t j = 0; j < size(); j++)
         _os << j << " " << aveDistances[j] << std::endl;
 }

◆ printSelf()

void CodeBook::printSelf ( std::ostream & _os ) const

virtual

Standard output for a code book Parameter: _os The output stream

Standard output for a codebook Parameter: _os The output stream

Reimplemented from ClassificationTrainingSet< FeatureVector, Label >.

Reimplemented in FuzzyMap, and ClassificationMap.

Definition at line 361 of file code_book.cpp.

 {
     ClassificationTrainingSet<FeatureVector, Label>::printSelf(_os);
 }

◆ readClassifVectors()

void CodeBook::readClassifVectors ( std::istream & _is )

protected

Reads the classif vectors from a stream. Parameter: _is The input stream

Definition at line 439 of file code_book.cpp.

 {
     int dim;
     _is >> dim;
     classifVectors.resize(dim);
     for (size_t i = 0; i < classifVectors.size(); i++)
         _is >> classifVectors[i];
 }

◆ readSelf()

void CodeBook::readSelf	(	std::istream &	_is,
		long	_dim = `-1`,
		long	_size = `-1`
	)

virtual

Standard input for a code book Parameter: _is The input stream

Standard input for a codebook Parameter: _is The input stream

Definition at line 370 of file code_book.cpp.

 {
 
 #ifndef _NO_EXCEPTION
     try
     {
 #endif
         clear();
         std::string line;
 
         // Determines the number of rows and columns in the training set
 
         long dim;
         long size;
         if (_dim == -1)
         {
             _is >> dim;
         }
         else
             dim = _dim;
         if (_size == -1)
         {
             _is >> line;
             if (!sscanf(line.c_str(), "%ld", &size))
             {
                 int x;
                 int y;
                 _is >> x;
                 _is >> y;
                 size = x * y;
             }
         }
         else
             size = _size;
         getline(_is, line);
         theItems.resize(size);
         theTargets.resize(size);
 
         for (int i = 0; i < size; i++)
         {
             std::vector<floatFeature> v;
             v.resize(dim);
             for (int j = 0; j < dim; j++)
             {
                 floatFeature var;
                 _is >> var;
                 v[j] = var;
             }
             getline(_is, line);
             theItems[i] = v;
             theTargets[i] = line;
         }
 #ifndef _NO_EXCEPTION
 
     }
     catch (std::exception& e)
     {
         std::ostringstream msg;
         msg << e.what() << std::endl << "Error reading the code book";
         throw std::runtime_error(msg.str());
     }
 #endif
 }

◆ saveObject()

void CodeBook::saveObject ( std::ostream & _os ) const

virtual

Saves the xmippCodeBook class into a stream. this method can be used to save the status of the class. Parameter: _os The output stream

Saves the CodeBook class into a stream. this method can be used to save the status of the class. Parameter: _os The output stream

Reimplemented from ClassificationTrainingSet< FeatureVector, Label >.

Reimplemented in FuzzyMap, ClassificationMap, and FuzzyCodeBook.

Definition at line 466 of file code_book.cpp.

 {
     writeClassifVectors(_os);
     ClassificationTrainingSet<FeatureVector, Label>::saveObject(_os);
 }

◆ test()

FeatureVector & CodeBook::test ( const FeatureVector & _in ) const

virtual

Returns the code vector that represents the input in the codebook Parameter: _in Sample to classify

Definition at line 172 of file code_book.cpp.

 {
     // eval the first one to init best & bestDist
     std::vector<FeatureVector>::const_iterator i = itemsBegin();
     std::vector<FeatureVector>::const_iterator best = i;
     double bestDist = euclideanDistance(*i, _in);
 
     // eval the rest
     for (i++ ; i < itemsEnd() ; i++)
     {
         double dist = euclideanDistance(*i, _in);
         if (dist < bestDist)
         {
             bestDist = dist;
             best = i;
         }
     }
 
     return (FeatureVector&)*best;
 }

◆ testIndex()

unsigned CodeBook::testIndex ( const FeatureVector & _in ) const

virtual

Returns the index to the code vector that represents the input in the codebook Parameter: _in Sample to classify

Definition at line 197 of file code_book.cpp.

 {
     // eval the first one to init best & bestDist
     std::vector<FeatureVector>::const_iterator i = itemsBegin();
     std::vector<FeatureVector>::const_iterator best = i;
     double bestDist = euclideanDistance(*i, _in);
 
     // eval the rest
     unsigned bestIndex = 0;
     unsigned index = 1;
     for (i++ ; i < itemsEnd() ; i++)
     {
         double dist = euclideanDistance(*i, _in);
         if (dist < bestDist)
         {
             bestDist = dist;
             best = i;
             bestIndex = index;
         }
         index++;
     }
 
     return bestIndex;
 }

◆ unNormalize()

void CodeBook::unNormalize ( const std::vector< ClassicTrainingVectors::statsStruct > & _varStats )

virtual

UnNormalize all features in the codebook Parameter: _varStats The normalization information

Definition at line 491 of file code_book.cpp.

 {
     using namespace std;
     if (_varStats.size() != theItems[0].size())
     {
         std::ostringstream msg;
         msg << "Normalization information does not coincide with codebook structure";
         throw std::runtime_error(msg.str());
     }
     for (unsigned it = 0; it < size(); it++)
     {
         for (unsigned i = 0; i < theItems[0].size(); i++)
         {
             if (!isnan(theItems[it][i]))
                 theItems[it][i] = theItems[it][i] * _varStats[i].sd + _varStats[i].mean;
         }
     }
 }

◆ writeClassifVectors()

void CodeBook::writeClassifVectors ( std::ostream & _os ) const

protected

Writes the classif vectors to a stream Parameter: _os The output stream

Definition at line 453 of file code_book.cpp.

 {
     _os << classifVectors.size() << std::endl;
     for (size_t i = 0; i < classifVectors.size(); i++)
         _os << classifVectors[i] << std::endl;
 }

Member Data Documentation

◆ aveDistances

std::vector< double > CodeBook::aveDistances

Definition at line 62 of file code_book.h.

◆ classifVectors

std::vector< std::vector <unsigned> > CodeBook::classifVectors

Definition at line 61 of file code_book.h.

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

xmipp/libraries/classification/code_book.h
xmipp/libraries/classification/code_book.cpp

Public Member Functions

Public Attributes

Protected Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ CodeBook() [1/5]

◆ CodeBook() [2/5]

◆ CodeBook() [3/5]

◆ CodeBook() [4/5]

◆ CodeBook() [5/5]

◆ ~CodeBook()

Member Function Documentation

◆ apply()

◆ calibrate()

◆ classifAt()

◆ classifSizeAt()

◆ classify()

◆ loadObject()

◆ Normalize()

◆ output()

◆ printHistogram()

◆ printQuantError()

◆ printSelf()

◆ readClassifVectors()

◆ readSelf()

◆ saveObject()

◆ test()

◆ testIndex()

◆ unNormalize()

◆ writeClassifVectors()

Member Data Documentation

◆ aveDistances

◆ classifVectors