freeling::adaboost Class Reference

Class AdaBoost implement a generic AB learner/classifier, which may be based on any kind of weak rule. More...

#include <adaboost.h>

Inheritance diagram for freeling::adaboost:

Collaboration diagram for freeling::adaboost:

List of all members.

Public Member Functions
	adaboost (int nl, std::wstring t)
	constructors, destructor and access methods
	adaboost (const std::wstring &file, const std::wstring &codes)
int	n_rules () const
void	classify (const example &i, double pred[]) const
	classification methods.
std::vector< double >	classify (const example &i) const
	classification returning vector: useful for Java API
void	pcl_ini_pointer ()
	partial classification
int	pcl_advance_pointer (int steps)
void	pcl_classify (const example &i, double *pred, int nrules)
	Important: pred is an array of predictions, one for each label the function *adds* its predicion for each label.
void	learn (dataset &ds, int nrounds, bool init, wr_params *p)
	learning methods
void	learn (dataset &ds, int nrounds, bool init, wr_params *p, const std::wstring &outf)
void	set_output (std::wostream *os)
	I/O methods.
void	read_from_stream (std::wistream *in)
void	read_from_file (const std::wstring &f)
void	set_initialize_weights (bool b)
Private Member Functions
void	initialize_weights (dataset &ds)
	auxiliar learning functions
void	update_weights (weak_rule *wr, double Z, dataset &ds)
void	add_weak_rule (weak_rule *wr)
	adaboost (const adaboost &old_bab)
	copy constructor forbidden
Private Attributes
bool	option_initialize_weights
	class parameters
std::wstring	wr_type
	type of used weak rules
adaboost::const_iterator	pcl_pointer
int	nrules
std::wostream *	out
	output

---

Detailed Description

Class AdaBoost implement a generic AB learner/classifier, which may be based on any kind of weak rule.

---

Constructor & Destructor Documentation

freeling::adaboost::adaboost

(

const adaboost &

old_bab

)

[private]

copy constructor forbidden

freeling::adaboost::adaboost	(	int	nl,
		std::wstring	t
	)

constructors, destructor and access methods

freeling::adaboost::adaboost	(	const std::wstring &	file,
		const std::wstring &	codes
	)

---

Member Function Documentation

void freeling::adaboost::add_weak_rule

(

weak_rule *

wr

)

[private]

void freeling::adaboost::classify	(	const example &	i,
		double	pred[]
	)		const [virtual]

classification methods.

Important: pred is an array of predictions, one for each label the function *assigns* its predicion for each label.

Implements freeling::classifier.

std::vector<double> freeling::adaboost::classify

(

const example &

i

)

const

classification returning vector: useful for Java API

void freeling::adaboost::initialize_weights

(

dataset &

ds

)

[private]

auxiliar learning functions

void freeling::adaboost::learn	(	dataset &	ds,
		int	nrounds,
		bool	init,
		wr_params *	p
	)

learning methods

void freeling::adaboost::learn	(	dataset &	ds,
		int	nrounds,
		bool	init,
		wr_params *	p,
		const std::wstring &	outf
	)

int freeling::adaboost::n_rules

(

)

const

int freeling::adaboost::pcl_advance_pointer

(

int

steps

)

void freeling::adaboost::pcl_classify	(	const example &	i,
		double *	pred,
		int	nrules
	)

Important: pred is an array of predictions, one for each label the function *adds* its predicion for each label.

void freeling::adaboost::pcl_ini_pointer

(

)

partial classification

void freeling::adaboost::read_from_file

(

const std::wstring &

f

)

void freeling::adaboost::read_from_stream

(

std::wistream *

in

)

void freeling::adaboost::set_initialize_weights

(

bool

b

)

void freeling::adaboost::set_output

(

std::wostream *

os

)

I/O methods.

void freeling::adaboost::update_weights	(	weak_rule *	wr,
		double	Z,
		dataset &	ds
	)		[private]

---

Member Data Documentation

int freeling::adaboost::nrules [private]

bool freeling::adaboost::option_initialize_weights [private]

class parameters

std::wostream* freeling::adaboost::out [private]

output

adaboost::const_iterator freeling::adaboost::pcl_pointer [private]

std::wstring freeling::adaboost::wr_type [private]

type of used weak rules

---

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

• adaboost.h