|
Daniel@0
|
1 <html>
|
|
Daniel@0
|
2 <head>
|
|
Daniel@0
|
3 <title>
|
|
Daniel@0
|
4 Netlab Reference Manual rbffwd
|
|
Daniel@0
|
5 </title>
|
|
Daniel@0
|
6 </head>
|
|
Daniel@0
|
7 <body>
|
|
Daniel@0
|
8 <H1> rbffwd
|
|
Daniel@0
|
9 </H1>
|
|
Daniel@0
|
10 <h2>
|
|
Daniel@0
|
11 Purpose
|
|
Daniel@0
|
12 </h2>
|
|
Daniel@0
|
13 Forward propagation through RBF network with linear outputs.
|
|
Daniel@0
|
14
|
|
Daniel@0
|
15 <p><h2>
|
|
Daniel@0
|
16 Synopsis
|
|
Daniel@0
|
17 </h2>
|
|
Daniel@0
|
18 <PRE>
|
|
Daniel@0
|
19 a = rbffwd(net, x)
|
|
Daniel@0
|
20 function [a, z, n2] = rbffwd(net, x)
|
|
Daniel@0
|
21 </PRE>
|
|
Daniel@0
|
22
|
|
Daniel@0
|
23
|
|
Daniel@0
|
24 <p><h2>
|
|
Daniel@0
|
25 Description
|
|
Daniel@0
|
26 </h2>
|
|
Daniel@0
|
27 <CODE>a = rbffwd(net, x)</CODE> takes a network data structure
|
|
Daniel@0
|
28 <CODE>net</CODE> and a matrix <CODE>x</CODE> of input
|
|
Daniel@0
|
29 vectors and forward propagates the inputs through the network to generate
|
|
Daniel@0
|
30 a matrix <CODE>a</CODE> of output vectors. Each row of <CODE>x</CODE> corresponds to one
|
|
Daniel@0
|
31 input vector and each row of <CODE>a</CODE> contains the corresponding output vector.
|
|
Daniel@0
|
32 The activation function that is used is determined by <CODE>net.actfn</CODE>.
|
|
Daniel@0
|
33
|
|
Daniel@0
|
34 <p><CODE>[a, z, n2] = rbffwd(net, x)</CODE> also generates a matrix <CODE>z</CODE> of
|
|
Daniel@0
|
35 the hidden unit activations where each row corresponds to one pattern.
|
|
Daniel@0
|
36 These hidden unit activations represent the <CODE>design matrix</CODE> for
|
|
Daniel@0
|
37 the RBF. The matrix <CODE>n2</CODE> is the squared distances between each
|
|
Daniel@0
|
38 basis function centre and each pattern in which each row corresponds
|
|
Daniel@0
|
39 to a data point.
|
|
Daniel@0
|
40
|
|
Daniel@0
|
41 <p><h2>
|
|
Daniel@0
|
42 Examples
|
|
Daniel@0
|
43 </h2>
|
|
Daniel@0
|
44 <PRE>
|
|
Daniel@0
|
45
|
|
Daniel@0
|
46 [a, z] = rbffwd(net, x);
|
|
Daniel@0
|
47
|
|
Daniel@0
|
48 <p>temp = pinv([z ones(size(x, 1), 1)]) * t;
|
|
Daniel@0
|
49 net.w2 = temp(1: nd(2), :);
|
|
Daniel@0
|
50 net.b2 = temp(size(x, nd(2)) + 1, :);
|
|
Daniel@0
|
51 </PRE>
|
|
Daniel@0
|
52
|
|
Daniel@0
|
53 Here <CODE>x</CODE> is the input data, <CODE>t</CODE> are the target values, and we use the
|
|
Daniel@0
|
54 pseudo-inverse to find the output weights and biases.
|
|
Daniel@0
|
55
|
|
Daniel@0
|
56 <p><h2>
|
|
Daniel@0
|
57 See Also
|
|
Daniel@0
|
58 </h2>
|
|
Daniel@0
|
59 <CODE><a href="rbf.htm">rbf</a></CODE>, <CODE><a href="rbferr.htm">rbferr</a></CODE>, <CODE><a href="rbfgrad.htm">rbfgrad</a></CODE>, <CODE><a href="rbfpak.htm">rbfpak</a></CODE>, <CODE><a href="rbftrain.htm">rbftrain</a></CODE>, <CODE><a href="rbfunpak.htm">rbfunpak</a></CODE><hr>
|
|
Daniel@0
|
60 <b>Pages:</b>
|
|
Daniel@0
|
61 <a href="index.htm">Index</a>
|
|
Daniel@0
|
62 <hr>
|
|
Daniel@0
|
63 <p>Copyright (c) Ian T Nabney (1996-9)
|
|
Daniel@0
|
64
|
|
Daniel@0
|
65
|
|
Daniel@0
|
66 </body>
|
|
Daniel@0
|
67 </html> |
