Functions to learn modes from subsets of data. More...

#include <vnl/vnl_vector.h>
#include <vnl/vnl_matrix.h>
#include <vcl_vector.h>

Functions
void	mcal_extract_mode (vcl_vector< vnl_vector< double > > &dv, const vcl_vector< unsigned > &elements_used, vnl_vector< double > &mode, double &var)
	Computes one mode from used elements of each `dv`.
void	mcal_extract_mode (vcl_vector< vnl_vector< double > > &dv, vnl_vector< double > &mode, double &var)
	Computes one mode by applying PCA to `dv`.
void	mcal_extract_modes (vcl_vector< vnl_vector< double > > &dv, const vcl_vector< vcl_vector< unsigned > > &elements_used, unsigned max_modes, double var_prop, vnl_matrix< double > &modes, vnl_vector< double > &mode_var)
	Compute modes and associated variance of supplied data.

Detailed Description

Functions to learn modes from subsets of data.

Author:: Tim Cootes

Definition in file mcal_extract_mode.h.

Function Documentation

void mcal_extract_mode	(	vcl_vector< vnl_vector< double > > &	dv,
		const vcl_vector< unsigned > &	used,
		vnl_vector< double > &	mode,
		double &	var
	)

Computes one mode from used elements of each dv.

Compute a mode which only uses a subset of the elements, all other elements are set to zero. Effectives computes the first eigenvector of the covariance matrix formed from selecting the used elements of each dv[i], ie dv[i][used[j]]. Resulting vector is returned as a full length vector (the same size as dv[i]).

Parameters:

dv	is assumed to be zero mean.

The contribution of this vector is removed from each dv, dv[i]-=mode*b, where b=dv[i].mode

Effectives computes the first eigenvector of the covariance matrix formed from selecting the used elements of each dv[i], ie dv[i][used[j]]. Resulting vector is returned as a full length vector (the same size as dv[i]).

The contribution of this vector is removed from each dv, dv[i]-=mode*b, where b=dv[i].mode

Definition at line 22 of file mcal_extract_mode.cxx.

void mcal_extract_mode	(	vcl_vector< vnl_vector< double > > &	dv,
		vnl_vector< double > &	mode,
		double &	var
	)

Computes one mode by applying PCA to dv.

Effectives computes the first eigenvector of the covariance matrix. The contribution of this vector is removed from each dv, dv[i]-=mode*b, where b=dv[i].mode

Definition at line 76 of file mcal_extract_mode.cxx.

void mcal_extract_modes	(	vcl_vector< vnl_vector< double > > &	dv,
		const vcl_vector< vcl_vector< unsigned > > &	used,
		unsigned	max_modes,
		double	var_prop,
		vnl_matrix< double > &	modes,
		vnl_vector< double > &	mode_var
	)

Compute modes and associated variance of supplied data.

Parameters:

elements_used[i]	indicates the set of elements to be used for mode i. Modes beyond `used.size()` will use all elements.
used[i]	indicates the set of elements to be used for mode i. Modes beyond `used.size()` will use all elements.

Definition at line 107 of file mcal_extract_mode.cxx.