Public Member Functions | Protected Member Functions | Protected Attributes

CGMNPSVM Class Reference


Detailed Description

Class GMNPSVM implements a one vs. rest MultiClass SVM.

It uses CGMNPLib for training (in true multiclass-SVM fashion).

Definition at line 24 of file GMNPSVM.h.

Inheritance diagram for CGMNPSVM:
Inheritance graph
[legend]

List of all members.

Public Member Functions

 CGMNPSVM ()
 CGMNPSVM (float64_t C, CKernel *k, CLabels *lab)
virtual ~CGMNPSVM ()
virtual EClassifierType get_classifier_type ()
float64_tget_basealphas_ptr (index_t *y, index_t *x)
virtual const char * get_name () const

Protected Member Functions

virtual bool train_machine (CFeatures *data=NULL)

Protected Attributes

float64_tm_basealphas
index_t m_basealphas_y
index_t m_basealphas_x

Constructor & Destructor Documentation

CGMNPSVM (  ) 

default constructor

Definition at line 23 of file GMNPSVM.cpp.

CGMNPSVM ( float64_t  C,
CKernel k,
CLabels lab 
)

constructor

Parameters:
C constant C
k kernel
lab labels

Definition at line 29 of file GMNPSVM.cpp.

~CGMNPSVM (  )  [virtual]

default destructor

Definition at line 35 of file GMNPSVM.cpp.


Member Function Documentation

float64_t * get_basealphas_ptr ( index_t y,
index_t x 
)

required for CMKLMulticlass constraint computation

Parameters:
y height of basealphas
x width of basealphas
Returns:
basealphas basealphas[k][j] is the alpha for class k and sample j which is untransformed compared to the alphas stored in CSVM* members

Definition at line 182 of file GMNPSVM.cpp.

virtual EClassifierType get_classifier_type (  )  [virtual]

get classifier type

Returns:
classifier type GMNPSVM

Reimplemented from CMachine.

Definition at line 47 of file GMNPSVM.h.

virtual const char* get_name ( void   )  const [virtual]
Returns:
object name

Reimplemented from CSVM.

Definition at line 61 of file GMNPSVM.h.

bool train_machine ( CFeatures data = NULL  )  [protected, virtual]

train SVM

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 CMachine.

Definition at line 51 of file GMNPSVM.cpp.


Member Data Documentation

float64_t* m_basealphas [protected]

required for CMKLMulticlass stores the untransformed alphas of this algorithm whereas CSVM* members stores a transformed version of it m_basealphas[k][j] is the alpha for class k and sample j

Definition at line 81 of file GMNPSVM.h.

index_t m_basealphas_x [protected]

base alphas x

Definition at line 85 of file GMNPSVM.h.

index_t m_basealphas_y [protected]

base alphas y

Definition at line 83 of file GMNPSVM.h.


The documentation for this class was generated from the following files:
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Defines

SHOGUN Machine Learning Toolbox - Documentation