1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
|
function [prob,a] = mdnprob(mixparams, t)
%MDNPROB Computes the data probability likelihood for an MDN mixture structure.
%
% Description
% PROB = MDNPROB(MIXPARAMS, T) computes the probability P(T) of each
% data vector in T under the Gaussian mixture model represented by the
% corresponding entries in MIXPARAMS. Each row of T represents a single
% vector.
%
% [PROB, A] = MDNPROB(MIXPARAMS, T) also computes the activations A
% (i.e. the probability P(T|J) of the data conditioned on each
% component density) for a Gaussian mixture model.
%
% See also
% MDNERR, MDNPOST
%
% Copyright (c) Ian T Nabney (1996-2001)
% David J Evans (1998)
% Check arguments for consistency
errstring = consist(mixparams, 'mdnmixes');
if ~isempty(errstring)
error(errstring);
end
ntarget = size(t, 1);
if ntarget ~= size(mixparams.centres, 1)
error('Number of targets does not match number of mixtures')
end
if size(t, 2) ~= mixparams.dim_target
error('Target dimension does not match mixture dimension')
end
dim_target = mixparams.dim_target;
ntarget = size(t, 1);
% Calculate squared norm matrix, of dimension (ndata, ncentres)
% vector (ntarget * ncentres)
dist2 = mdndist2(mixparams, t);
% Calculate variance factors
variance = 2.*mixparams.covars;
% Compute the normalisation term
normal = ((2.*pi).*mixparams.covars).^(dim_target./2);
% Now compute the activations
a = exp(-(dist2./variance))./normal;
% Accumulate negative log likelihood of targets
prob = mixparams.mixcoeffs.*a;
|
