CScatterSVM Class Reference

---

Detailed Description

ScatterSVM - Multiclass SVM.

The ScatterSVM is an unpublished experimental true multiclass SVM. Details are availabe in the following technical report.

This code is currently experimental.

Robert Jenssen and Marius Kloft and Alexander Zien and S"oren Sonnenburg and Klaus-Robert M"{u}ller, A Multi-Class Support Vector Machine Based on Scatter Criteria, TR 014-2009 TU Berlin, 2009

Definition at line 50 of file ScatterSVM.h.

Inheritance diagram for CScatterSVM:

List of all members.

Public Member Functions
	CScatterSVM (void)
	CScatterSVM (SCATTER_TYPE type)
	CScatterSVM (float64_t C, CKernel *k, CLabels *lab)
virtual	~CScatterSVM ()
virtual bool	train (CFeatures *data=NULL)
virtual EClassifierType	get_classifier_type ()
virtual float64_t	classify_example (int32_t num)
virtual CLabels *	classify_one_vs_rest ()
virtual const char *	get_name () const
Protected Attributes
SCATTER_TYPE	scatter_type
svm_problem	problem
svm_parameter	param
struct svm_model *	model
float64_t *	norm_wc
float64_t *	norm_wcw
float64_t	rho
int32_t	m_num_classes

---

Constructor & Destructor Documentation

CScatterSVM

(

void

)

default constructor

Definition at line 20 of file ScatterSVM.cpp.

CScatterSVM

(

SCATTER_TYPE

type

)

constructor

Definition at line 27 of file ScatterSVM.cpp.

CScatterSVM	(	float64_t	C,
		CKernel *	k,
		CLabels *	lab
	)

constructor (using NO_BIAS as default scatter_type)

Parameters:

C	constant C
k	kernel
lab	labels

Definition at line 33 of file ScatterSVM.cpp.

~CScatterSVM

(

)

[virtual]

default destructor

Definition at line 39 of file ScatterSVM.cpp.

---

Member Function Documentation

float64_t classify_example

(

int32_t

num

)

[virtual]

classify one example

Parameters:

num	number of example to classify

Returns:

resulting classification

Reimplemented from CMultiClassSVM.

Definition at line 447 of file ScatterSVM.cpp.

CLabels * classify_one_vs_rest

(

)

[virtual]

classify one vs rest

Returns:

resulting labels

Reimplemented from CMultiClassSVM.

Definition at line 339 of file ScatterSVM.cpp.

virtual EClassifierType get_classifier_type

(

)

[virtual]

get classifier type

Returns:

classifier type LIBSVM

Reimplemented from CClassifier.

Definition at line 84 of file ScatterSVM.h.

virtual const char* get_name

(

void

)

const [virtual]

Returns:

object name

Reimplemented from CSVM.

Definition at line 100 of file ScatterSVM.h.

bool train

(

CFeatures *

data = NULL

)

[virtual]

train SVM classifier

Parameters:

data	training data (parameter can be avoided if distance or kernel-based classifiers are used and distance/kernels are initialized with train data)

Returns:

whether training was successful

Reimplemented from CClassifier.

Definition at line 45 of file ScatterSVM.cpp.

---

Member Data Documentation

int32_t m_num_classes [protected]

number of classes

Reimplemented from CMultiClassSVM.

Definition at line 130 of file ScatterSVM.h.

struct svm_model* model [protected]

SVM model

Definition at line 118 of file ScatterSVM.h.

float64_t* norm_wc [protected]

norm of w_c

Definition at line 121 of file ScatterSVM.h.

float64_t* norm_wcw [protected]

norm of w_cw

Definition at line 124 of file ScatterSVM.h.

svm_parameter param [protected]

SVM param

Definition at line 115 of file ScatterSVM.h.

svm_problem problem [protected]

SVM problem

Definition at line 113 of file ScatterSVM.h.

float64_t rho [protected]

ScatterSVM rho

Definition at line 127 of file ScatterSVM.h.

SCATTER_TYPE scatter_type [protected]

type of scatter SVM

Definition at line 110 of file ScatterSVM.h.

---

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

• ScatterSVM.h
• ScatterSVM.cpp