comparison toolboxes/FullBNT-1.0.7/netlab3.3/kmeansNetlab.m @ 0:e9a9cd732c1e tip

first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:e9a9cd732c1e
1 function [centres, options, post, errlog] = kmeans(centres, data, options)
2 %KMEANS Trains a k means cluster model.
3 %
4 % Description
5 % CENTRES = KMEANS(CENTRES, DATA, OPTIONS) uses the batch K-means
6 % algorithm to set the centres of a cluster model. The matrix DATA
7 % represents the data which is being clustered, with each row
8 % corresponding to a vector. The sum of squares error function is used.
9 % The point at which a local minimum is achieved is returned as
10 % CENTRES. The error value at that point is returned in OPTIONS(8).
11 %
12 % [CENTRES, OPTIONS, POST, ERRLOG] = KMEANS(CENTRES, DATA, OPTIONS)
13 % also returns the cluster number (in a one-of-N encoding) for each
14 % data point in POST and a log of the error values after each cycle in
15 % ERRLOG. The optional parameters have the following
16 % interpretations.
17 %
18 % OPTIONS(1) is set to 1 to display error values; also logs error
19 % values in the return argument ERRLOG. If OPTIONS(1) is set to 0, then
20 % only warning messages are displayed. If OPTIONS(1) is -1, then
21 % nothing is displayed.
22 %
23 % OPTIONS(2) is a measure of the absolute precision required for the
24 % value of CENTRES at the solution. If the absolute difference between
25 % the values of CENTRES between two successive steps is less than
26 % OPTIONS(2), then this condition is satisfied.
27 %
28 % OPTIONS(3) is a measure of the precision required of the error
29 % function at the solution. If the absolute difference between the
30 % error functions between two successive steps is less than OPTIONS(3),
31 % then this condition is satisfied. Both this and the previous
32 % condition must be satisfied for termination.
33 %
34 % OPTIONS(14) is the maximum number of iterations; default 100.
35 %
36 % See also
37 % GMMINIT, GMMEM
38 %
39
40 % Copyright (c) Ian T Nabney (1996-2001)
41
42 [ndata, data_dim] = size(data);
43 [ncentres, dim] = size(centres);
44
45 if dim ~= data_dim
46 error('Data dimension does not match dimension of centres')
47 end
48
49 if (ncentres > ndata)
50 error('More centres than data')
51 end
52
53 % Sort out the options
54 if (options(14))
55 niters = options(14);
56 else
57 niters = 100;
58 end
59
60 store = 0;
61 if (nargout > 3)
62 store = 1;
63 errlog = zeros(1, niters);
64 end
65
66 % Check if centres and posteriors need to be initialised from data
67 if (options(5) == 1)
68 % Do the initialisation
69 perm = randperm(ndata);
70 perm = perm(1:ncentres);
71
72 % Assign first ncentres (permuted) data points as centres
73 centres = data(perm, :);
74 end
75 % Matrix to make unit vectors easy to construct
76 id = eye(ncentres);
77
78 % Main loop of algorithm
79 for n = 1:niters
80
81 % Save old centres to check for termination
82 old_centres = centres;
83
84 % Calculate posteriors based on existing centres
85 d2 = dist2(data, centres);
86 % Assign each point to nearest centre
87 [minvals, index] = min(d2', [], 1);
88 post = id(index,:);
89
90 num_points = sum(post, 1);
91 % Adjust the centres based on new posteriors
92 for j = 1:ncentres
93 if (num_points(j) > 0)
94 centres(j,:) = sum(data(find(post(:,j)),:), 1)/num_points(j);
95 end
96 end
97
98 % Error value is total squared distance from cluster centres
99 e = sum(minvals);
100 if store
101 errlog(n) = e;
102 end
103 if options(1) > 0
104 fprintf(1, 'Cycle %4d Error %11.6f\n', n, e);
105 end
106
107 if n > 1
108 % Test for termination
109 if max(max(abs(centres - old_centres))) < options(2) & ...
110 abs(old_e - e) < options(3)
111 options(8) = e;
112 return;
113 end
114 end
115 old_e = e;
116 end
117
118 % If we get here, then we haven't terminated in the given number of
119 % iterations.
120 options(8) = e;
121 if (options(1) >= 0)
122 disp(maxitmess);
123 end
124