DecisionStump< DataType > Struct Template Reference

A Weak Classifier based on a Feature Extractor policy. More...

#include <DecisionStump.h>

Public Types
typedef int	ReturnType

Public Member Functions
int	evaluate_feature (DataType value) const
void	rescale (int sx, int sy, int sgn)
	this function permits to change the th for rescaling

Static Public Member Functions
static std::string	signature ()

Public Attributes
int	parity
	parity for this classifier (-1, 1)
DataType	th
	threshold value

Static Public Attributes
static const ClassifierType	Type = DicreteClassifier
	DecisionStump is a Discrete Classifier.

Detailed Description

template<class DataType>
struct DecisionStump< DataType >

A Weak Classifier based on a Feature Extractor policy.

The FeatureExtractor have to implement a response method that returns an integer scalar quantity parity=+1 -> class = (value < th) ? 1 : -1; parity=-1 -> class = (value > th) ? 1 : -1;

Note

parity=1 and parity=-1 are not exact opposite cases due to missing of =

value=th is an ambiguos case

Todo:

try to compact BoostableClassifier with DecisionStump in order to reduce number of multiplication (parity * alpha) (e.g. using RealDecisionStump classes)

Member Function Documentation

template<class DataType>

int DecisionStump< DataType >::evaluate_feature ( DataType  value ) const

inline

convert the feature value to {-1,+1} using internal parity and threshold parity=1 -> value < th ? 1 : -1 parity=-1 -> value > th ? 1 : -1

---

The documentation for this struct was generated from the following file:

• include/Classifier/DecisionStump.h