Main Page   Compound List   File List   Compound Members   File Members   Related Pages  

MI Class Reference

Object performing the analytical calculus of mutual information beween 3 dimensionnal Gaussians in consecutive times. More...

#include <MI.hh>

List of all members.

Public Methods

 MI ()
virtual ~MI ()
void Allocate (int nb_dimensions)
 Memory allocation. More...

void Initialise1 (vector< float ** > *C1, vector< float * > *M1)
 Initialize variable at T1 and calculates the covariance matrix determinant. More...

void Initialise2 (vector< float ** > *C2, vector< float * > *M2)
 Initialize variable at T2 and calculates the covariance matrix determinant. More...

void Initialise12 (vector< float ** > *C12, vector< float * > *M12)
 Initialize variable at T12 and calculates the covariance matrix determinant. More...

float MICalculus ()
 Mutual information calculation. More...


Private Methods

void DetCovMatrixT1 ()
 determinant calculation at T1. More...

void DetCovMatrixT2 ()
 determinant calculation at T2. More...

void DetCovMatrixT12 ()
 determinant calculation at T12. More...


Private Attributes

vector< float * > MeanT1
 mean vector time 1.

vector< float * > MeanT2
 mean vector time 2.

vector< float * > MeanT12
 mean vector time 1 and 2.

vector< float ** > CovT1
 covariance matrix time 1.

vector< float ** > CovT2
 covariance matrix time 2.

vector< float ** > CovT12
 covariance matrix time 1 and 2.

vector< float > determinantT1
 Determinant covariance matrix time 1.

vector< float > determinantT2
 Determinant covariance matrix time 2.

vector< float > determinantT12
 Determinant covariance matrix time 1 and 2.

int nb_dim
 dimensionnality.


Detailed Description

Object performing the analytical calculus of mutual information beween 3 dimensionnal Gaussians in consecutive times.

The procedure is based on the determinants of covaraint matrices corresponding to the 2 Gaussian distributions (T1 and T2) and to the joint distribution T12

Definition at line 14 of file MI.hh.


Member Function Documentation

void MI::Allocate int    nb_dimensions
 

Memory allocation.

This method should be simpler. Indeed, the vector objects used here could be replace by a simple scalar.

Parameters:
nb_dimensions  dimensionnality of the TL feature spaces

Definition at line 21 of file MI.cpp.

void MI::DetCovMatrixT1   [private]
 

determinant calculation at T1.

method based on a LU decomposition provided by Numerical recepies

Note that when the matrix is singular the determinants is calculated with a covariance matrix where the singular dimensions have been removed.

Definition at line 85 of file MI.cpp.

void MI::DetCovMatrixT12   [private]
 

determinant calculation at T12.

method based on a LU decomposition provided by Numerical recepies

Note that when the matrix is singular the determinants is calculated with a covariance matrix where the singular dimensions have been removed.

Definition at line 232 of file MI.cpp.

void MI::DetCovMatrixT2   [private]
 

determinant calculation at T2.

method based on a LU decomposition provided by Numerical recepies

Note that when the matrix is singular the determinants is calculated with a covariance matrix where the singular dimensions have been removed.

Definition at line 157 of file MI.cpp.

void MI::Initialise1 vector< float ** > *    C1,
vector< float * > *    M1
 

Initialize variable at T1 and calculates the covariance matrix determinant.

Parameters:
C1  covariance matrix of the MT cluster at time 1
M1  mean vector of the MT cluster at time 1

Definition at line 59 of file MI.cpp.

void MI::Initialise12 vector< float ** > *    C12,
vector< float * > *    M12
 

Initialize variable at T12 and calculates the covariance matrix determinant.

Parameters:
C12  covariance matrix of the MT cluster projected at time 1 and 2
M12  mean vector of the MT cluster projected at time 1 and 2

Definition at line 77 of file MI.cpp.

void MI::Initialise2 vector< float ** > *    C2,
vector< float * > *    M2
 

Initialize variable at T2 and calculates the covariance matrix determinant.

Parameters:
C2  covariance matrix of the MT cluster at time 1
M2  mean vector of the MT cluster at time 1

Definition at line 68 of file MI.cpp.

float MI::MICalculus  
 

Mutual information calculation.

This function computes a simple ration of determinants for the mutual information calculation. Note that if there are singular covariance matrices, the mutual information is still equal to the ratio of determinants, but the singular dimensions must be removed from the marginal and joint covariance matrices.

Returns:
mutual information value

Definition at line 310 of file MI.cpp.


The documentation for this class was generated from the following files:
Generated on Thu Feb 17 11:01:55 2005 for Inference of a Graph of Dynamic Cluster Trajectories by doxygen1.2.14 written by Dimitri van Heesch, © 1997-2002