|
Daniel@0
|
1 <html>
|
|
Daniel@0
|
2 <head>
|
|
Daniel@0
|
3 <title>
|
|
Daniel@0
|
4 Netlab Reference Manual kmeans
|
|
Daniel@0
|
5 </title>
|
|
Daniel@0
|
6 </head>
|
|
Daniel@0
|
7 <body>
|
|
Daniel@0
|
8 <H1> kmeans
|
|
Daniel@0
|
9 </H1>
|
|
Daniel@0
|
10 <h2>
|
|
Daniel@0
|
11 Purpose
|
|
Daniel@0
|
12 </h2>
|
|
Daniel@0
|
13 Trains a k means cluster model.
|
|
Daniel@0
|
14
|
|
Daniel@0
|
15 <p><h2>
|
|
Daniel@0
|
16 Synopsis
|
|
Daniel@0
|
17 </h2>
|
|
Daniel@0
|
18 <PRE>
|
|
Daniel@0
|
19 centres = kmeans(centres, data, options)
|
|
Daniel@0
|
20 [centres, options] = kmeans(centres, data, options)
|
|
Daniel@0
|
21 [centres, options, post, errlog] = kmeans(centres, data, options)
|
|
Daniel@0
|
22 </PRE>
|
|
Daniel@0
|
23
|
|
Daniel@0
|
24
|
|
Daniel@0
|
25 <p><h2>
|
|
Daniel@0
|
26 Description
|
|
Daniel@0
|
27 </h2>
|
|
Daniel@0
|
28
|
|
Daniel@0
|
29 <CODE>centres = kmeans(centres, data, options)</CODE>
|
|
Daniel@0
|
30 uses the batch K-means algorithm to set the centres of a cluster model.
|
|
Daniel@0
|
31 The matrix <CODE>data</CODE> represents the data
|
|
Daniel@0
|
32 which is being clustered, with each row corresponding to a vector.
|
|
Daniel@0
|
33 The sum of squares error function is used. The point at which
|
|
Daniel@0
|
34 a local minimum is achieved is returned as <CODE>centres</CODE>. The
|
|
Daniel@0
|
35 error value at that point is returned in <CODE>options(8)</CODE>.
|
|
Daniel@0
|
36
|
|
Daniel@0
|
37 <p><CODE>[centres, options, post, errlog] = kmeans(centres, data, options)</CODE>
|
|
Daniel@0
|
38 also returns the cluster number (in a one-of-N encoding) for each data
|
|
Daniel@0
|
39 point in <CODE>post</CODE> and a log of the error values after each cycle in
|
|
Daniel@0
|
40 <CODE>errlog</CODE>.
|
|
Daniel@0
|
41
|
|
Daniel@0
|
42 The optional parameters have the following interpretations.
|
|
Daniel@0
|
43
|
|
Daniel@0
|
44 <p><CODE>options(1)</CODE> is set to 1 to display error values; also logs error
|
|
Daniel@0
|
45 values in the return argument <CODE>errlog</CODE>.
|
|
Daniel@0
|
46 If <CODE>options(1)</CODE> is set to 0,
|
|
Daniel@0
|
47 then only warning messages are displayed. If <CODE>options(1)</CODE> is -1,
|
|
Daniel@0
|
48 then nothing is displayed.
|
|
Daniel@0
|
49
|
|
Daniel@0
|
50 <p><CODE>options(2)</CODE> is a measure of the absolute precision required for the value
|
|
Daniel@0
|
51 of <CODE>centres</CODE> at the solution. If the absolute difference between
|
|
Daniel@0
|
52 the values of <CODE>centres</CODE> between two successive steps is less than
|
|
Daniel@0
|
53 <CODE>options(2)</CODE>, then this condition is satisfied.
|
|
Daniel@0
|
54
|
|
Daniel@0
|
55 <p><CODE>options(3)</CODE> is a measure of the precision required of the error
|
|
Daniel@0
|
56 function at the solution. If the absolute difference between the
|
|
Daniel@0
|
57 error functions between two successive steps is less than
|
|
Daniel@0
|
58 <CODE>options(3)</CODE>, then this condition is satisfied.
|
|
Daniel@0
|
59 Both this and the previous condition must be
|
|
Daniel@0
|
60 satisfied for termination.
|
|
Daniel@0
|
61
|
|
Daniel@0
|
62 <p><CODE>options(14)</CODE> is the maximum number of iterations; default 100.
|
|
Daniel@0
|
63
|
|
Daniel@0
|
64 <p><h2>
|
|
Daniel@0
|
65 Example
|
|
Daniel@0
|
66 </h2>
|
|
Daniel@0
|
67 <CODE>kmeans</CODE> can be used to initialise the centres of a Gaussian
|
|
Daniel@0
|
68 mixture model that is then trained with the EM algorithm.
|
|
Daniel@0
|
69 <PRE>
|
|
Daniel@0
|
70
|
|
Daniel@0
|
71 [priors, centres, var] = gmmunpak(p, md);
|
|
Daniel@0
|
72 centres = kmeans(centres, data, options);
|
|
Daniel@0
|
73 p = gmmpak(priors, centres, var);
|
|
Daniel@0
|
74 p = gmmem(p, md, data, options);
|
|
Daniel@0
|
75 </PRE>
|
|
Daniel@0
|
76
|
|
Daniel@0
|
77
|
|
Daniel@0
|
78 <p><h2>
|
|
Daniel@0
|
79 See Also
|
|
Daniel@0
|
80 </h2>
|
|
Daniel@0
|
81 <CODE><a href="gmminit.htm">gmminit</a></CODE>, <CODE><a href="gmmem.htm">gmmem</a></CODE><hr>
|
|
Daniel@0
|
82 <b>Pages:</b>
|
|
Daniel@0
|
83 <a href="index.htm">Index</a>
|
|
Daniel@0
|
84 <hr>
|
|
Daniel@0
|
85 <p>Copyright (c) Ian T Nabney (1996-9)
|
|
Daniel@0
|
86
|
|
Daniel@0
|
87
|
|
Daniel@0
|
88 </body>
|
|
Daniel@0
|
89 </html> |
