Mercurial > hg > camir-aes2014

comparison toolboxes/FullBNT-1.0.7/nethelp3.3/mdn.htm @ 0:e9a9cd732c1e tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
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 <html>
2 <head>
3 <title>
4 Netlab Reference Manual mdn
5 </title>
6 </head>
7 <body>
8 <H1> mdn
9 </H1>
10 <h2>
11 Purpose
12 </h2>
13 Creates a Mixture Density Network with specified architecture.
14
15 <p><h2>
16 Synopsis
17 </h2>
18 <PRE>
19 net = mdn(nin, nhidden, ncentres, dimtarget)
20 net = mdn(nin, nhidden, ncentres, dimtarget, mixtype, ...
21 prior, beta)
22 </PRE>
23
24
25 <p><h2>
26 Description
27 </h2>
28 <CODE>net = mdn(nin, nhidden, ncentres, dimtarget)</CODE> takes the number of
29 inputs,
30 hidden units for a 2-layer feed-forward
31 network and the number of centres and target dimension for the
32 mixture model whose parameters are set from the outputs of the neural network.
33 The fifth argument <CODE>mixtype</CODE> is used to define the type of mixture
34 model. (Currently there is only one type supported: a mixture of Gaussians with
35 a single covariance parameter for each component.) For this model,
36 the mixture coefficients are computed from a group of softmax outputs,
37 the centres are equal to a group of linear outputs, and the variances are
38 obtained by applying the exponential function to a third group of outputs.
39
40 <p>The network is initialised by a call to <CODE>mlp</CODE>, and the arguments
41 <CODE>prior</CODE>, and <CODE>beta</CODE> have the same role as for that function.
42 Weight initialisation uses the Matlab function <CODE>randn</CODE>
43 and so the seed for the random weight initialization can be
44 set using <CODE>randn('state', s)</CODE> where <CODE>s</CODE> is the seed value.
45 A specialised data structure (rather than <CODE>gmm</CODE>)
46 is used for the mixture model outputs to improve
47 the efficiency of error and gradient calculations in network training.
48 The fields are described in <CODE>mdnfwd</CODE> where they are set up.
49
50 <p>The fields in <CODE>net</CODE> are
51 <PRE>
52
53 type = 'mdn'
54 nin = number of input variables
55 nout = dimension of target space (not number of network outputs)
56 nwts = total number of weights and biases
57 mdnmixes = data structure for mixture model output
58 mlp = data structure for MLP network
59 </PRE>
60
61
62 <p><h2>
63 Example
64 </h2>
65 <PRE>
66
67 net = mdn(2, 4, 3, 1, 'spherical');
68 </PRE>
69
70 This creates a Mixture Density Network with 2 inputs and 4 hidden units.
71 The mixture model has 3 components and the target space has dimension 1.
72
73 <p><h2>
74 See Also
75 </h2>
76 <CODE><a href="mdnfwd.htm">mdnfwd</a></CODE>, <CODE><a href="mdnerr.htm">mdnerr</a></CODE>, <CODE><a href="mdn2gmm.htm">mdn2gmm</a></CODE>, <CODE><a href="mdngrad.htm">mdngrad</a></CODE>, <CODE><a href="mdnpak.htm">mdnpak</a></CODE>, <CODE><a href="mdnunpak.htm">mdnunpak</a></CODE>, <CODE><a href="mlp.htm">mlp</a></CODE><hr>
77 <b>Pages:</b>
78 <a href="index.htm">Index</a>
79 <hr>
80 <p>Copyright (c) Ian T Nabney (1996-9)
81 <p>David J Evans (1998)
82
83 </body>
84 </html>