Programmer Guide/Command Reference/EVAL/em - Revision history

193.171.195.48 at 06:19, 21 April 2011

2011-04-21T06:19:28Z

← Older revision		Revision as of 08:19, 21 April 2011
Line 14:		Line 14:
	::*if '''1''', the accurate log-likelihood estimation is used		::*if '''1''', the accurate log-likelihood estimation is used
	::*otherwise use the Intel IPP fast log. add method		::*otherwise use the Intel IPP fast log. add method
	;Result: The estimated model parameters are stored in the references ''meantable'', ''sigmatable'', ''wtable'' and (optional) ''aposterioritable''. The return value ~~is 10 element vector~~ ''r'' containing informations about the estimation process.		;Result: The estimated model parameters are stored in the references ''meantable'', ''sigmatable'', ''wtable'' and (optional) ''aposterioritable''. The return value ''r'' is vector with 10 elements, containing informations about the estimation process.
	::{\|class="einrahmen"		::{\|class="einrahmen"
	\|''r''[0] \|\|number of feature vectors		\|''r''[0] \|\|number of feature vectors

193.171.195.48 at 06:18, 21 April 2011

2011-04-21T06:18:23Z

← Older revision		Revision as of 08:18, 21 April 2011
Line 34:		Line 34:
	\|''r''[8] \|\|value of the "accurate" parameter		\|''r''[8] \|\|value of the "accurate" parameter
	\|-		\|-
	\|''r''[9] \|\|'''0'' (= method used is EM for diagonal covariance matrices)		\|''r''[9] \|\|'''0''' (= method used is EM for diagonal covariance matrices)
	\|}		\|}
	;See also: [[../haclust\|haclust]], [[../modclust\|modclust]], [[../density\|density]], [[../svd\|svd]]		;See also: [[../haclust\|haclust]], [[../modclust\|modclust]], [[../density\|density]], [[../svd\|svd]]

	[[../#Functions\|<function list>]]		[[../#Functions\|<function list>]]

193.171.195.48 at 06:17, 21 April 2011

2011-04-21T06:17:56Z

← Older revision		Revision as of 08:17, 21 April 2011
Line 17:		Line 17:
	::{\|class="einrahmen"		::{\|class="einrahmen"
	\|''r''[0] \|\|number of feature vectors		\|''r''[0] \|\|number of feature vectors
	\|''r''[0] \|\|number of dimensions		\|-
	\|''r''[0] \|\|number of mixture components		\|''r''[1] \|\|number of dimensions
	\|''r''[0] \|\|number of iterations performed by the algorithm		\|-
	\|''r''[0] \|\|joint log-likelihood of the data in the estimated model		\|''r''[2] \|\|number of mixture components
	\|''r''[0] \|\|'''-1''' (initialisation max. iterations, apparently unused)		\|-
	\|''r''[0] \|\|maximum number of iterations		\|''r''[3] \|\|number of iterations performed by the algorithm
	\|''r''[0] \|\|log-likelihood threshold		\|-
	\|''r''[0] \|\|value of the "accurate" parameter		\|''r''[4] \|\|joint log-likelihood of the data in the estimated model
	\|''r''[0] \|\|'''0'' (= method used is EM for diagonal covariance matrices)		\|-
			\|''r''[5] \|\|'''-1''' (initialisation max. iterations, apparently unused)
			\|-
			\|''r''[6] \|\|maximum number of iterations
			\|-
			\|''r''[7] \|\|log-likelihood threshold
			\|-
			\|''r''[8] \|\|value of the "accurate" parameter
			\|-
			\|''r''[9] \|\|'''0'' (= method used is EM for diagonal covariance matrices)
	\|}		\|}
	;See also: [[../haclust\|haclust]], [[../modclust\|modclust]], [[../density\|density]], [[../svd\|svd]]		;See also: [[../haclust\|haclust]], [[../modclust\|modclust]], [[../density\|density]], [[../svd\|svd]]

	[[../#Functions\|<function list>]]		[[../#Functions\|<function list>]]

193.171.195.48: Created page with '{{DISPLAYTITLE:{{SUBPAGENAME}}}} Implementation of the Expectation Maximization (EM) algorithm for finding maximum likelihood estimates of the parameters of a Gaussian Mixture Mo…'

2011-04-21T06:16:55Z

Created page with '{{DISPLAYTITLE:{{SUBPAGENAME}}}} Implementation of the Expectation Maximization (EM) algorithm for finding maximum likelihood estimates of the parameters of a Gaussian Mixture Mo…'

New page

{{DISPLAYTITLE:{{SUBPAGENAME}}}}
Implementation of the Expectation Maximization (EM) algorithm for finding maximum likelihood estimates of the parameters of a Gaussian Mixture Model (GMM) (i.e. estimates the model parameters for which the observed data are most likely).
;Usage:<code>em(<var>x</var>, <var>meantable</var>, <var>membertable</var>, <var>sigmatable</var>, <var>wtable</var>, <var>aposterioritable</var>, <var>minvar</var>, <var>emmaxit</var>, <var>likthresh</var> {, <var>accurate</var>})</code>
:;<var>x</var>:feature vectors, 1 feature vector per row ((Nvectors x Ndimensions)
:;<var>meantable</var>:output matrix for estimated mean vectors (Nmixtures x Ndimensions)
:;<var>membertable</var>:vector with Nvectors elements, defining the feature vector memberships to mixtures, resulting from prior clustering
:;<var>sigmatable</var>:output matrix for estimated covariance vectors (i.e. diagonal of covariance matrices) of mixture model (Nmixtures x Ndimensions)
:;<var>wtable</var>:output vector of estimated mixture weights (Nmixtures)
:;<var>aposterioritable</var>:if a table item is given, it will contain the output a posteriori probabilities of the feature vectors, otherwise (if a number is given) it won't be changed
:;<var>minvar</var>:minimum variance
:;<var>emmaxit</var>:maximum number of iterations
:;<var>liktresh</var>:log-likelihood threshold; if the difference between log-likelihoods of two consecutive iterations is less than this threshold, the algorithm halts
:;<var>accurate</var>:
::*if '''1''', the accurate log-likelihood estimation is used
::*otherwise use the Intel IPP fast log. add method
;Result: The estimated model parameters are stored in the references ''meantable'', ''sigmatable'', ''wtable'' and (optional) ''aposterioritable''. The return value is 10 element vector ''r'' containing informations about the estimation process.
::{|class="einrahmen"
|''r''[0] ||number of feature vectors
|''r''[0] ||number of dimensions
|''r''[0] ||number of mixture components
|''r''[0] ||number of iterations performed by the algorithm
|''r''[0] ||joint log-likelihood of the data in the estimated model
|''r''[0] ||'''-1''' (initialisation max. iterations, apparently unused)
|''r''[0] ||maximum number of iterations
|''r''[0] ||log-likelihood threshold
|''r''[0] ||value of the "accurate" parameter
|''r''[0] ||'''0'' (= method used is EM for diagonal covariance matrices)
|}
;See also: [[../haclust|haclust]], [[../modclust|modclust]], [[../density|density]], [[../svd|svd]]

[[../#Functions|<function list>]]