Mercurial > hg > camir-aes2014

comparison toolboxes/FullBNT-1.0.7/nethelp3.3/mlpprior.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 mlpprior
5 </title>
6 </head>
7 <body>
8 <H1> mlpprior
9 </H1>
10 <h2>
11 Purpose
12 </h2>
13 Create Gaussian prior for mlp.
14
15 <p><h2>
16 Synopsis
17 </h2>
18 <PRE>
19 prior = mlpprior(nin, nhidden, nout, aw1, ab1, aw2, ab2)</PRE>
20
21
22 <p><h2>
23 Description
24 </h2>
25 <CODE>prior = mlpprior(nin, nhidden, nout, aw1, ab1, aw2, ab2)</CODE>
26 generates a data structure
27 <CODE>prior</CODE>, with fields <CODE>prior.alpha</CODE> and <CODE>prior.index</CODE>, which
28 specifies a Gaussian prior distribution for the network weights in a
29 two-layer feedforward network. Two different cases are possible. In
30 the first case, <CODE>aw1</CODE>, <CODE>ab1</CODE>, <CODE>aw2</CODE> and <CODE>ab2</CODE> are all
31 scalars and represent the regularization coefficients for four groups
32 of parameters in the network corresponding to first-layer weights,
33 first-layer biases, second-layer weights, and second-layer biases
34 respectively. Then <CODE>prior.alpha</CODE> represents a column vector of
35 length 4 containing the parameters, and <CODE>prior.index</CODE> is a matrix
36 specifying which weights belong in each group. Each column has one
37 element for each weight in the matrix, using the standard ordering as
38 defined in <CODE>mlppak</CODE>, and each element is 1 or 0 according to
39 whether the weight is a member of the corresponding group or not. In
40 the second case the parameter <CODE>aw1</CODE> is a vector of length equal to
41 the number of inputs in the network, and the corresponding matrix
42 <CODE>prior.index</CODE> now partitions the first-layer weights into groups
43 corresponding to the weights fanning out of each input unit. This
44 prior is appropriate for the technique of automatic relevance
45 determination.
46
47 <p><h2>
48 See Also
49 </h2>
50 <CODE><a href="mlp.htm">mlp</a></CODE>, <CODE><a href="mlperr.htm">mlperr</a></CODE>, <CODE><a href="mlpgrad.htm">mlpgrad</a></CODE>, <CODE><a href="evidence.htm">evidence</a></CODE><hr>
51 <b>Pages:</b>
52 <a href="index.htm">Index</a>
53 <hr>
54 <p>Copyright (c) Ian T Nabney (1996-9)
55
56
57 </body>
58 </html>