Class GMNPSVM implements a one vs. rest MultiClass SVM.
It uses CGMNPLib for training (in true multiclass-SVM fashion).
Definition at line 26 of file GMNPSVM.h.
Public Member Functions | |
CGMNPSVM () | |
CGMNPSVM (float64_t C, CKernel *k, CLabels *lab) | |
virtual | ~CGMNPSVM () |
virtual bool | train (CFeatures *data=NULL) |
virtual EClassifierType | get_classifier_type () |
void | getbasealphas (::std::vector< ::std::vector< float64_t > > &basealphas) |
virtual const char * | get_name () const |
Protected Attributes | |
::std::vector< ::std::vector < float64_t > > | m_basealphas |
CGMNPSVM | ( | ) |
default constructor
Definition at line 23 of file GMNPSVM.cpp.
~CGMNPSVM | ( | ) | [virtual] |
default destructor
Definition at line 33 of file GMNPSVM.cpp.
virtual EClassifierType get_classifier_type | ( | ) | [virtual] |
virtual const char* get_name | ( | ) | const [virtual] |
void getbasealphas | ( | ::std::vector< ::std::vector< float64_t > > & | basealphas | ) |
required for CMKLMulticlass constraint computation
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 163 of file GMNPSVM.cpp.
bool train | ( | CFeatures * | data = NULL |
) | [virtual] |
train SVM
data | training data (parameter can be avoided if distance or kernel-based classifiers are used and distance/kernels are initialized with train data) |
Reimplemented from CClassifier.
Definition at line 37 of file GMNPSVM.cpp.
::std::vector< ::std::vector<float64_t> > m_basealphas [protected] |