|
wolffd@0
|
1 <html>
|
|
wolffd@0
|
2 <head>
|
|
wolffd@0
|
3 <title>
|
|
wolffd@0
|
4 Netlab Reference Manual netopt
|
|
wolffd@0
|
5 </title>
|
|
wolffd@0
|
6 </head>
|
|
wolffd@0
|
7 <body>
|
|
wolffd@0
|
8 <H1> netopt
|
|
wolffd@0
|
9 </H1>
|
|
wolffd@0
|
10 <h2>
|
|
wolffd@0
|
11 Purpose
|
|
wolffd@0
|
12 </h2>
|
|
wolffd@0
|
13 Optimize the weights in a network model.
|
|
wolffd@0
|
14
|
|
wolffd@0
|
15 <p><h2>
|
|
wolffd@0
|
16 Synopsis
|
|
wolffd@0
|
17 </h2>
|
|
wolffd@0
|
18 <PRE>
|
|
wolffd@0
|
19 [net, options] = netopt(net, options, x, t, alg)
|
|
wolffd@0
|
20 [net, options, varargout] = netopt(net, options, x, t, alg)
|
|
wolffd@0
|
21 </PRE>
|
|
wolffd@0
|
22
|
|
wolffd@0
|
23
|
|
wolffd@0
|
24 <p><h2>
|
|
wolffd@0
|
25 Description
|
|
wolffd@0
|
26 </h2>
|
|
wolffd@0
|
27
|
|
wolffd@0
|
28 <p><CODE>netopt</CODE> is a helper function which facilitates the training of
|
|
wolffd@0
|
29 networks using the general purpose optimizers as well as sampling from the
|
|
wolffd@0
|
30 posterior distribution of parameters using general purpose Markov chain
|
|
wolffd@0
|
31 Monte Carlo sampling algorithms. It can be used with any function that
|
|
wolffd@0
|
32 searches in parameter space using error and gradient functions.
|
|
wolffd@0
|
33
|
|
wolffd@0
|
34 <p><CODE>[net, options] = netopt(net, options, x, t, alg)</CODE> takes a network
|
|
wolffd@0
|
35 data structure <CODE>net</CODE>, together with a vector <CODE>options</CODE> of
|
|
wolffd@0
|
36 parameters governing the behaviour of the optimization algorithm, a
|
|
wolffd@0
|
37 matrix <CODE>x</CODE> of input vectors and a matrix <CODE>t</CODE> of target
|
|
wolffd@0
|
38 vectors, and returns the trained network as well as an updated
|
|
wolffd@0
|
39 <CODE>options</CODE> vector. The string <CODE>alg</CODE> determines which optimization
|
|
wolffd@0
|
40 algorithm (<CODE>conjgrad</CODE>, <CODE>quasinew</CODE>, <CODE>scg</CODE>, etc.) or Monte
|
|
wolffd@0
|
41 Carlo algorithm (such as <CODE>hmc</CODE>) will be used.
|
|
wolffd@0
|
42
|
|
wolffd@0
|
43 <p><CODE>[net, options, varargout] = netopt(net, options, x, t, alg)</CODE>
|
|
wolffd@0
|
44 also returns any additional return values from the optimisation algorithm.
|
|
wolffd@0
|
45
|
|
wolffd@0
|
46 <p><h2>
|
|
wolffd@0
|
47 Examples
|
|
wolffd@0
|
48 </h2>
|
|
wolffd@0
|
49 Suppose we create a 4-input, 3 hidden unit, 2-output feed-forward
|
|
wolffd@0
|
50 network using <CODE>net = mlp(4, 3, 2, 'linear')</CODE>. We can then train
|
|
wolffd@0
|
51 the network with the scaled conjugate gradient algorithm by using
|
|
wolffd@0
|
52 <CODE>net = netopt(net, options, x, t, 'scg')</CODE> where <CODE>x</CODE> and
|
|
wolffd@0
|
53 <CODE>t</CODE> are the input and target data matrices respectively, and the
|
|
wolffd@0
|
54 options vector is set appropriately for <CODE>scg</CODE>.
|
|
wolffd@0
|
55
|
|
wolffd@0
|
56 <p>If we also wish to plot the learning curve, we can use the additional
|
|
wolffd@0
|
57 return value <CODE>errlog</CODE> given by <CODE>scg</CODE>:
|
|
wolffd@0
|
58 <PRE>
|
|
wolffd@0
|
59
|
|
wolffd@0
|
60 [net, options, errlog] = netopt(net, options, x, t, 'scg');
|
|
wolffd@0
|
61 </PRE>
|
|
wolffd@0
|
62
|
|
wolffd@0
|
63
|
|
wolffd@0
|
64 <p><h2>
|
|
wolffd@0
|
65 See Also
|
|
wolffd@0
|
66 </h2>
|
|
wolffd@0
|
67 <CODE><a href="netgrad.htm">netgrad</a></CODE>, <CODE><a href="bfgs.htm">bfgs</a></CODE>, <CODE><a href="conjgrad.htm">conjgrad</a></CODE>, <CODE><a href="graddesc.htm">graddesc</a></CODE>, <CODE><a href="hmc.htm">hmc</a></CODE>, <CODE><a href="scg.htm">scg</a></CODE><hr>
|
|
wolffd@0
|
68 <b>Pages:</b>
|
|
wolffd@0
|
69 <a href="index.htm">Index</a>
|
|
wolffd@0
|
70 <hr>
|
|
wolffd@0
|
71 <p>Copyright (c) Ian T Nabney (1996-9)
|
|
wolffd@0
|
72
|
|
wolffd@0
|
73
|
|
wolffd@0
|
74 </body>
|
|
wolffd@0
|
75 </html> |
