annotate toolboxes/FullBNT-1.0.7/nethelp3.3/gmm.htm @ 0:e9a9cd732c1e tip

first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
rev   line source
wolffd@0 1 <html>
wolffd@0 2 <head>
wolffd@0 3 <title>
wolffd@0 4 Netlab Reference Manual gmm
wolffd@0 5 </title>
wolffd@0 6 </head>
wolffd@0 7 <body>
wolffd@0 8 <H1> gmm
wolffd@0 9 </H1>
wolffd@0 10 <h2>
wolffd@0 11 Purpose
wolffd@0 12 </h2>
wolffd@0 13 Creates a Gaussian mixture model with specified architecture.
wolffd@0 14
wolffd@0 15 <p><h2>
wolffd@0 16 Synopsis
wolffd@0 17 </h2>
wolffd@0 18 <PRE>
wolffd@0 19 mix = gmm(dim, ncentres, covartype)
wolffd@0 20 mix = gmm(dim, ncentres, covartype, ppca_dim)
wolffd@0 21 </PRE>
wolffd@0 22
wolffd@0 23
wolffd@0 24 <p><h2>
wolffd@0 25 Description
wolffd@0 26 </h2>
wolffd@0 27
wolffd@0 28 <CODE>mix = gmm(dim, ncentres, covartype)</CODE> takes
wolffd@0 29 the dimension of the space <CODE>dim</CODE>, the number of centres in the
wolffd@0 30 mixture model and the type of the mixture model, and returns a data
wolffd@0 31 structure <CODE>mix</CODE>.
wolffd@0 32 The mixture model type defines the covariance structure of each component
wolffd@0 33 Gaussian:
wolffd@0 34 <PRE>
wolffd@0 35
wolffd@0 36 'spherical' = single variance parameter for each component: stored as a vector
wolffd@0 37 'diag' = diagonal matrix for each component: stored as rows of a matrix
wolffd@0 38 'full' = full matrix for each component: stored as 3d array
wolffd@0 39 'ppca' = probabilistic PCA: stored as principal components (in a 3d array
wolffd@0 40 and associated variances and off-subspace noise
wolffd@0 41 </PRE>
wolffd@0 42
wolffd@0 43 <CODE>mix = gmm(dim, ncentres, covartype, ppca_dim)</CODE> also sets the dimension of
wolffd@0 44 the PPCA sub-spaces: the default value is one.
wolffd@0 45
wolffd@0 46 <p>The priors are initialised to equal values summing to one, and the covariances
wolffd@0 47 are all the identity matrix (or equivalent). The centres are
wolffd@0 48 initialised randomly from a zero mean unit variance Gaussian. This makes use
wolffd@0 49 of the MATLAB function <CODE>randn</CODE> and so the seed for the random weight
wolffd@0 50 initialisation can be set using <CODE>randn('state', s)</CODE> where <CODE>s</CODE> is the
wolffd@0 51 state value.
wolffd@0 52
wolffd@0 53 <p>The fields in <CODE>mix</CODE> are
wolffd@0 54 <PRE>
wolffd@0 55
wolffd@0 56 type = 'gmm'
wolffd@0 57 nin = the dimension of the space
wolffd@0 58 ncentres = number of mixture components
wolffd@0 59 covartype = string for type of variance model
wolffd@0 60 priors = mixing coefficients
wolffd@0 61 centres = means of Gaussians: stored as rows of a matrix
wolffd@0 62 covars = covariances of Gaussians
wolffd@0 63 </PRE>
wolffd@0 64
wolffd@0 65 The additional fields for mixtures of PPCA are
wolffd@0 66 <PRE>
wolffd@0 67
wolffd@0 68 U = principal component subspaces
wolffd@0 69 lambda = in-space covariances: stored as rows of a matrix
wolffd@0 70 </PRE>
wolffd@0 71
wolffd@0 72 The off-subspace noise is stored in <CODE>covars</CODE>.
wolffd@0 73
wolffd@0 74 <p><h2>
wolffd@0 75 Example
wolffd@0 76 </h2>
wolffd@0 77 <PRE>
wolffd@0 78
wolffd@0 79 mix = gmm(2, 4, 'spherical');
wolffd@0 80 </PRE>
wolffd@0 81
wolffd@0 82 This creates a Gaussian mixture model with 4 components in 2 dimensions.
wolffd@0 83 The covariance structure is a spherical model.
wolffd@0 84
wolffd@0 85 <p><h2>
wolffd@0 86 See Also
wolffd@0 87 </h2>
wolffd@0 88 <CODE><a href="gmmpak.htm">gmmpak</a></CODE>, <CODE><a href="gmmunpak.htm">gmmunpak</a></CODE>, <CODE><a href="gmmsamp.htm">gmmsamp</a></CODE>, <CODE><a href="gmminit.htm">gmminit</a></CODE>, <CODE><a href="gmmem.htm">gmmem</a></CODE>, <CODE><a href="gmmactiv.htm">gmmactiv</a></CODE>, <CODE><a href="gmmpost.htm">gmmpost</a></CODE>, <CODE><a href="gmmprob.htm">gmmprob</a></CODE><hr>
wolffd@0 89 <b>Pages:</b>
wolffd@0 90 <a href="index.htm">Index</a>
wolffd@0 91 <hr>
wolffd@0 92 <p>Copyright (c) Ian T Nabney (1996-9)
wolffd@0 93
wolffd@0 94
wolffd@0 95 </body>
wolffd@0 96 </html>