GestureRecognitionToolkit  Version: 1.0 Revision: 04-03-15
The Gesture Recognition Toolkit (GRT) is a cross-platform, open-source, c++ machine learning library for real-time gesture recognition.
GRT::DecisionStump Class Reference
Inheritance diagram for GRT::DecisionStump:
GRT::WeakClassifier

Public Member Functions

 DecisionStump (const UINT numRandomSplits=100)
 
virtual ~DecisionStump ()
 
 DecisionStump (const DecisionStump &rhs)
 
DecisionStumpoperator= (const DecisionStump &rhs)
 
virtual bool deepCopyFrom (const WeakClassifier *weakClassifer)
 
virtual bool train (ClassificationData &trainingData, VectorDouble &weights)
 
virtual double predict (const VectorDouble &x)
 
virtual bool saveModelToFile (fstream &file) const
 
virtual bool loadModelFromFile (fstream &file)
 
virtual void print () const
 
UINT getDecisionFeatureIndex () const
 
UINT getDirection () const
 
UINT getNumRandomSplits () const
 
double getDecisionValue () const
 
- Public Member Functions inherited from GRT::WeakClassifier
 WeakClassifier ()
 
virtual ~WeakClassifier ()
 
 WeakClassifier (const WeakClassifier &rhs)
 
WeakClassifieroperator= (const WeakClassifier &rhs)
 
bool copyBaseVariables (const WeakClassifier *weakClassifer)
 
virtual double getPositiveClassLabel () const
 
virtual double getNegativeClassLabel () const
 
string getWeakClassifierType () const
 
bool getTrained () const
 
UINT getNumInputDimensions () const
 
WeakClassifiercreateNewInstance () const
 

Protected Attributes

UINT decisionFeatureIndex
 The dimension that the data will be spilt on.
 
UINT direction
 Indicates if the decision spilt threshold is greater than (1), or less than (0)
 
UINT numRandomSplits
 The number of random splits used to search for the best decision spilt.
 
double decisionValue
 The decision spilt threshold.
 
- Protected Attributes inherited from GRT::WeakClassifier
string weakClassifierType
 A string that represents the weak classifier type, e.g. DecisionStump.
 
bool trained
 A flag to show if the weak classifier model has been trained.
 
UINT numInputDimensions
 The number of input dimensions to the weak classifier.
 
TrainingLog trainingLog
 
ErrorLog errorLog
 
WarningLog warningLog
 

Static Protected Attributes

static RegisterWeakClassifierModule< DecisionStumpregisterModule
 This is used to register the DecisionStump with the WeakClassifier base class.
 

Additional Inherited Members

- Public Types inherited from GRT::WeakClassifier
typedef std::map< string, WeakClassifier *(*)() > StringWeakClassifierMap
 
- Static Public Member Functions inherited from GRT::WeakClassifier
static WeakClassifiercreateInstanceFromString (string const &weakClassifierType)
 
- Static Protected Member Functions inherited from GRT::WeakClassifier
static StringWeakClassifierMapgetMap ()
 

Detailed Description

Definition at line 36 of file DecisionStump.h.

Constructor & Destructor Documentation

DecisionStump::DecisionStump ( const UINT  numRandomSplits = 100)

Default Constructor.

Sets the number of random splits that will be used to search for the best split value.

Parameters
constUINT numRandomSplits: sets the number of random splits that will be used to search for the best split value. Default value = 100

Definition at line 35 of file DecisionStump.cpp.

DecisionStump::~DecisionStump ( )
virtual

Default Destructor.

Definition at line 48 of file DecisionStump.cpp.

DecisionStump::DecisionStump ( const DecisionStump rhs)

Default Copy Constructor. Defines how the data from the rhs GRT::DecisionStump instance is copied to this GRT::DecisionStump instance.

Definition at line 52 of file DecisionStump.cpp.

Member Function Documentation

bool DecisionStump::deepCopyFrom ( const WeakClassifier weakClassifer)
virtual

This function enables the data from one GRT::DecisionStump instance to be copied into this GRT::DecisionStump instance.

Parameters
WeakClassifier*weakClassifer: a pointer to the Classifier Base Class, this should be pointing to another GRT::DecisionStump instance
Returns
returns true if the clone was successfull, false otherwise

Reimplemented from GRT::WeakClassifier.

Definition at line 67 of file DecisionStump.cpp.

UINT DecisionStump::getDecisionFeatureIndex ( ) const
Returns
returns the index of the feature that is used to split the data into a positive or negative class

Definition at line 255 of file DecisionStump.cpp.

double DecisionStump::getDecisionValue ( ) const
Returns
returns the decision spilt threshold

Definition at line 267 of file DecisionStump.cpp.

UINT DecisionStump::getDirection ( ) const
Returns
returns if the decision spilt threshold is greater than (1), or less than (0)

Definition at line 259 of file DecisionStump.cpp.

UINT DecisionStump::getNumRandomSplits ( ) const
Returns
returns the number of random splits that will be used to search for the best decision spilt

Definition at line 263 of file DecisionStump.cpp.

bool DecisionStump::loadModelFromFile ( fstream &  file)
virtual

This function loads an model model from a file.

fstream &file: a reference to the file you want to load the RBF model from

Returns
returns true if the model was loaded successfull, false otherwise

Reimplemented from GRT::WeakClassifier.

Definition at line 180 of file DecisionStump.cpp.

DecisionStump & DecisionStump::operator= ( const DecisionStump rhs)

Defines how the data from the rhs GRT::DecisionStump instance is copied to this GRT::DecisionStump instance.

Definition at line 56 of file DecisionStump.cpp.

double DecisionStump::predict ( const VectorDouble &  x)
virtual

This function predicts the class label of the input vector, given the current model. The class label returned will either be positive (WEAK_CLASSIFIER_POSITIVE_CLASS_LABEL) or negative (WEAK_CLASSIFIER_NEGATIVE_CLASS_LABEL).

Parameters
VectorDouble&weights: a reference to the vector used for prediction
Returns
returns the estimated class label, which will be positive or negative

Reimplemented from GRT::WeakClassifier.

Definition at line 150 of file DecisionStump.cpp.

void DecisionStump::print ( ) const
virtual

This function prints out some basic info about the model to std::cout.

Reimplemented from GRT::WeakClassifier.

Definition at line 248 of file DecisionStump.cpp.

bool DecisionStump::saveModelToFile ( fstream &  file) const
virtual

This function saves the current model to a file.

fstream &file: a reference to the file you want to save the RBF model to

Returns
returns true if the model was saved successfull, false otherwise

Reimplemented from GRT::WeakClassifier.

Definition at line 157 of file DecisionStump.cpp.

bool DecisionStump::train ( ClassificationData trainingData,
VectorDouble &  weights 
)
virtual

This function trains the DecisionStump model, using the weighted labelled training data.

Parameters
ClassificationData&trainingData: the labelled training data
VectorDouble&weights: the corresponding weights for each sample in the labelled training data
Returns
returns true if the model was trained successfull, false otherwise

Reimplemented from GRT::WeakClassifier.

Definition at line 77 of file DecisionStump.cpp.


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