Estats Class Reference

Computes statistics between 2 DT for HEM E-step. More...

#include <Estats.hpp>

List of all members.

Public Member Functions

 Estats (std::vector< Dytex > &dti, std::vector< Dytex > &dtj, int tau, bool useYmean)
 initialize Estats object.
void saveCache ()
 RUN AND CACHE THE SENS ANALYSIS FOR DTJ.
void computeEll ()
 Compute Expected log-likelihood between base and reduced DTs.
void computeAggregateStats (Mat W)
 Save aggregate statistics for reduced DTs only.
double logdetiid (Mat Q, double r, int C)
 compute logdet of transformed covariance w/ diagonal or iid noise.

Public Attributes

std::vector< Dytex > dti
 cell array of DT components of base mixture
std::vector< Dytex > dtj
 cell array of DT component of reduced mixture
int tau
 sequence length
bool useYmean
 1 = use Ymean, 0 = don't use Ymean
Mat Ell
 Expected log-likelihood between Dts.
std::vector< Mat > xij
 aggregate statistic.
std::vector< Mat > etaj
 aggregate statistic.
std::vector< Mat > Phij
 aggregate statistic.
std::vector< Mat > varphij
 aggregate statistic.
std::vector< Mat > phij
 aggregate statistic.
std::vector< Mat > betaj
 aggregate statistic.
std::vector< Mat > Psij
 aggregate statistic.
std::vector< Mat > Gammaj
 aggregate statistic.
std::vector< double > Lambdaj
 aggregate statistic.
std::vector< Mat > gammaj
 aggregate statistic.
std::vector< Mat > dtjsa_Q_Ft
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_Q_GtC2
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_Q_Wt
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_Q_foo
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_Q_Jt
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_Q_Ht
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_Lt
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_LF
 SENS ANALYSIS Cache variable.
std::vector< Mat > dtjsa_LtGt
 SENS ANALYSIS Cache variable.
Mat dtjsa_iA2
 SENS ANALYSIS Cache variable.
Mat dtjsa_Szero
 SENS ANALYSIS Cache variable.
Mat dtjsa_Psij_init
 SENS ANALYSIS Cache variable.
Mat dtjsa_etaj_init
 SENS ANALYSIS Cache variable.
Mat dtjsa_varphij_init
 SENS ANALYSIS Cache variable.
Mat dtjsa_Phij_init
 SENS ANALYSIS Cache variable.
Mat dtjsa_phij_init
 SENS ANALYSIS Cache variable.
Mat dtjsa_Q_logdet
 SENS ANALYSIS Cache variable.
std::vector< bool > dtjblank
 SENS ANALYSIS Cache variable.

Detailed Description

Computes statistics between 2 DT for HEM E-step.

Write detailed description for Estats here.

Remarks:
Fatser version and takes less memory because only aggregate statistics are precomputed and stored
See also:
DytexMix::runHEM

Definition at line 26 of file Estats.hpp.


Constructor & Destructor Documentation

Estats::Estats ( std::vector< Dytex > &  dti,
std::vector< Dytex > &  dtj,
int  tau,
bool  useYmean 
)

initialize Estats object.

Parameters:
dti cell array of DT components of base mixture.
dtj cell array of DT component of reduced mixture.
tau sequence length.
useYmean 1 = use Ymean, 0 = don't use Ymean.

initialize SENS ANALYSIS Cache for base and reduced DTs

Definition at line 51 of file Estats.cpp.

Here is the call graph for this function:


Member Function Documentation

void Estats::computeAggregateStats ( Mat  W  ) 

Save aggregate statistics for reduced DTs only.

Parameters:
W normalize weight matrix
See also:
computeEll.

Definition at line 715 of file Estats.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

double Estats::logdetiid ( Mat  Q,
double  r,
int  C 
)

compute logdet of transformed covariance w/ diagonal or iid noise.

Parameters:
Q parameter Q.
r parameter r.
C parameter C.
Returns:
logdet value.
See also:
saveCache.

Definition at line 1266 of file Estats.cpp.

Here is the caller graph for this function:


Member Data Documentation

std::vector<bool> Estats::dtjblank

SENS ANALYSIS Cache variable.

check if reduced DT is blank or not

Definition at line 206 of file Estats.hpp.


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

Generated on 18 Dec 2014 for libdt-OpenCV by  doxygen 1.6.1