class LDATrainer

This class is designed to compute Linear Discriminant Analysis

Inheritance:


Public Fields

[more]int n_inputs
number of inputs
[more]int n_classes
number of classes
[more]Mat* Sb
between-class scatter matrix Sb of size n_inputsxn_inputs
[more]Mat* Sw
within-class scatter matrix Sw of size n_inputsxn_inputs
[more]real** mean_
means of each class
[more]real* mean_class_mean_
Mean of class means
[more]real* d_m_mean_
projection of mean_class_mean_ into LDA space
[more]Vec* eigenvalues
eigenvalues in descending order of
[more]real* eigenvalues_
eigenvalues of the LDAMachine (to be filled)
[more]Mat* eigenvectors
LDA basis vector
[more]real** eigenvectors_
LDA basis vector of the LDAMachine (to be filled)
[more]int verbose_level
verbosity level
[more]bool classprovided
consider the target of an example as its class, otherwise attribute the class assuming the same number of example per class
[more]bool within_is_identity
consider the within scatter matrix as the identity matrix
[more]bool save
save scatter matrices

Public Methods

[more] LDATrainer(int n_classes_, LDAMachine* machine_)
[more]void makeSbSw(DataSet* data_)
compute between and within scatter matrices
[more]virtual void train(DataSet* data_, MeasurerList* measurers_ = NULL) = 0
[more]virtual ~LDATrainer()


Documentation

This class is designed to compute Linear Discriminant Analysis

This class only computes the between-class scatter matrix and the within-class scatter matrix.

oint n_inputs
number of inputs

oint n_classes
number of classes

oMat* Sb
between-class scatter matrix Sb of size n_inputsxn_inputs

oMat* Sw
within-class scatter matrix Sw of size n_inputsxn_inputs

oreal** mean_
means of each class

oreal* mean_class_mean_
Mean of class means

oreal* d_m_mean_
projection of mean_class_mean_ into LDA space

oVec* eigenvalues
eigenvalues in descending order of

oreal* eigenvalues_
eigenvalues of the LDAMachine (to be filled)

oMat* eigenvectors
LDA basis vector

oreal** eigenvectors_
LDA basis vector of the LDAMachine (to be filled)

oint verbose_level
verbosity level

obool classprovided
consider the target of an example as its class, otherwise attribute the class assuming the same number of example per class

obool within_is_identity
consider the within scatter matrix as the identity matrix

obool save
save scatter matrices

o LDATrainer(int n_classes_, LDAMachine* machine_)

ovoid makeSbSw(DataSet* data_)
compute between and within scatter matrices

ovirtual void train(DataSet* data_, MeasurerList* measurers_ = NULL) = 0

ovirtual ~LDATrainer()


Direct child classes:
DirectLDATrainer
Author:
Sebastien Marcel (marcel@idiap.ch) Fri 15 Jul 2005 11:35:27 AM CEST
Version:
2.0
Since:
1.0

Alphabetic index HTML hierarchy of classes or Java



This page was generated with the help of DOC++.