← Back to branch summary

~ubuntu-branches/ubuntu/lucid/python-scipy/lucid

« back to all changes in this revision

Viewing changes to Lib/sandbox/ann/doc/srn.html

  • Committer: Bazaar Package Importer
  • Author(s): Matthias Klose
  • Date: 2007-01-07 14:12:12 UTC
  • mfrom: (1.1.1 upstream)
  • Revision ID: james.westby@ubuntu.com-20070107141212-mm0ebkh5b37hcpzn
http://bugs.debian.org/402783
* Remove build dependency on python-numpy-dev.
* python-scipy: Depend on python-numpy instead of python-numpy-dev.
* Package builds on other archs than i386. Closes: #402783.

Show diffs side-by-side

added added

removed removed

Lines of Context:
Lib/sandbox/ann/doc/srn.html
 
1
 
 
2
<!doctype html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
 
3
<html><head><title>Python: module srn</title>
 
4
</head><body bgcolor="#f0f0f8">
 
5
 
 
6
<table width="100%" cellspacing=0 cellpadding=2 border=0 summary="heading">
 
7
<tr bgcolor="#7799ee">
 
8
<td valign=bottom>&nbsp;<br>
 
9
<font color="#ffffff" face="helvetica, arial">&nbsp;<br><big><big><strong>srn</strong></big></big></font></td
 
10
><td align=right valign=bottom
 
11
><font color="#ffffff" face="helvetica, arial"><a href=".">index</a><br><a href="file:/home/tilde/programming/SoC/scipy/Lib/sandbox/ann/srn.py">/home/tilde/programming/SoC/scipy/Lib/sandbox/ann/srn.py</a></font></td></tr></table>
 
12
    <p><tt>#&nbsp;<a href="#srn">srn</a>.py<br>
 
13
#&nbsp;by:&nbsp;Fred&nbsp;Mailhot<br>
 
14
#&nbsp;last&nbsp;mod:&nbsp;2006-08-18</tt></p>
 
15
<p>
 
16
<table width="100%" cellspacing=0 cellpadding=2 border=0 summary="section">
 
17
<tr bgcolor="#aa55cc">
 
18
<td colspan=3 valign=bottom>&nbsp;<br>
 
19
<font color="#fffff" face="helvetica, arial"><big><strong>Modules</strong></big></font></td></tr>
 
20
    
 
21
<tr><td bgcolor="#aa55cc"><tt>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</tt></td><td>&nbsp;</td>
 
22
<td width="100%"><table width="100%" summary="list"><tr><td width="25%" valign=top><a href="numpy.html">numpy</a><br>
 
23
</td><td width="25%" valign=top></td><td width="25%" valign=top></td><td width="25%" valign=top></td></tr></table></td></tr></table><p>
 
24
<table width="100%" cellspacing=0 cellpadding=2 border=0 summary="section">
 
25
<tr bgcolor="#ee77aa">
 
26
<td colspan=3 valign=bottom>&nbsp;<br>
 
27
<font color="#ffffff" face="helvetica, arial"><big><strong>Classes</strong></big></font></td></tr>
 
28
    
 
29
<tr><td bgcolor="#ee77aa"><tt>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</tt></td><td>&nbsp;</td>
 
30
<td width="100%"><dl>
 
31
<dt><font face="helvetica, arial"><a href="srn.html#srn">srn</a>
 
32
</font></dt></dl>
 
33
 <p>
 
34
<table width="100%" cellspacing=0 cellpadding=2 border=0 summary="section">
 
35
<tr bgcolor="#ffc8d8">
 
36
<td colspan=3 valign=bottom>&nbsp;<br>
 
37
<font color="#000000" face="helvetica, arial"><a name="srn">class <strong>srn</strong></a></font></td></tr>
 
38
    
 
39
<tr bgcolor="#ffc8d8"><td rowspan=2><tt>&nbsp;&nbsp;&nbsp;</tt></td>
 
40
<td colspan=2><tt>Class&nbsp;to&nbsp;define,&nbsp;train&nbsp;and&nbsp;test&nbsp;a&nbsp;simple&nbsp;recurrent&nbsp;network<br>&nbsp;</tt></td></tr>
 
41
<tr><td>&nbsp;</td>
 
42
<td width="100%">Methods defined here:<br>
 
43
<dl><dt><a name="srn-__init__"><strong>__init__</strong></a>(self, ni, nh, no, f<font color="#909090">='linear'</font>, w<font color="#909090">=None</font>)</dt><dd><tt>Set&nbsp;up&nbsp;instance&nbsp;of&nbsp;<a href="#srn">srn</a>.&nbsp;Initial&nbsp;weights&nbsp;are&nbsp;drawn&nbsp;from&nbsp;a&nbsp;<br>
 
44
zero-mean&nbsp;Gaussian&nbsp;w/&nbsp;variance&nbsp;is&nbsp;scaled&nbsp;by&nbsp;fan-in.<br>
 
45
Input:<br>
 
46
&nbsp;&nbsp;&nbsp;&nbsp;ni&nbsp;&nbsp;-&nbsp;&lt;int&gt;&nbsp;#&nbsp;of&nbsp;inputs<br>
 
47
&nbsp;&nbsp;&nbsp;&nbsp;nh&nbsp;&nbsp;-&nbsp;&lt;int&gt;&nbsp;#&nbsp;of&nbsp;hidden&nbsp;&amp;&nbsp;context&nbsp;units<br>
 
48
&nbsp;&nbsp;&nbsp;&nbsp;no&nbsp;&nbsp;-&nbsp;&lt;int&gt;&nbsp;#&nbsp;of&nbsp;outputs<br>
 
49
&nbsp;&nbsp;&nbsp;&nbsp;f&nbsp;&nbsp;&nbsp;-&nbsp;&lt;str&gt;&nbsp;output&nbsp;activation&nbsp;fxn<br>
 
50
&nbsp;&nbsp;&nbsp;&nbsp;w&nbsp;&nbsp;&nbsp;-&nbsp;&lt;array&nbsp;dtype=Float&gt;&nbsp;weight&nbsp;vector</tt></dd></dl>
 
51
 
 
52
<dl><dt><a name="srn-errfxn"><strong>errfxn</strong></a>(self, w, x, t)</dt><dd><tt>Return&nbsp;vector&nbsp;of&nbsp;squared-errors&nbsp;for&nbsp;the&nbsp;leastsq&nbsp;optimizer</tt></dd></dl>
 
53
 
 
54
<dl><dt><a name="srn-fwd_all"><strong>fwd_all</strong></a>(self, x, w<font color="#909090">=None</font>)</dt><dd><tt>Propagate&nbsp;values&nbsp;forward&nbsp;through&nbsp;the&nbsp;net.&nbsp;<br>
 
55
Input:<br>
 
56
&nbsp;&nbsp;&nbsp;&nbsp;x&nbsp;&nbsp;&nbsp;-&nbsp;matrix&nbsp;of&nbsp;all&nbsp;input&nbsp;patterns<br>
 
57
&nbsp;&nbsp;&nbsp;&nbsp;w&nbsp;&nbsp;&nbsp;-&nbsp;1-d&nbsp;vector&nbsp;of&nbsp;weights<br>
 
58
Returns:<br>
 
59
&nbsp;&nbsp;&nbsp;&nbsp;y&nbsp;&nbsp;&nbsp;-&nbsp;matrix&nbsp;of&nbsp;all&nbsp;outputs</tt></dd></dl>
 
60
 
 
61
<dl><dt><a name="srn-pack"><strong>pack</strong></a>(self)</dt><dd><tt>Compile&nbsp;weight&nbsp;matrices&nbsp;w1,b1,wc,w2,b2&nbsp;from&nbsp;net&nbsp;into&nbsp;a<br>
 
62
single&nbsp;vector,&nbsp;suitable&nbsp;for&nbsp;optimization&nbsp;routines.</tt></dd></dl>
 
63
 
 
64
<dl><dt><a name="srn-test_all"><strong>test_all</strong></a>(self, x, t)</dt><dd><tt>Test&nbsp;network&nbsp;on&nbsp;an&nbsp;array&nbsp;(size&gt;1)&nbsp;of&nbsp;patterns<br>
 
65
Input:<br>
 
66
&nbsp;&nbsp;&nbsp;&nbsp;x&nbsp;&nbsp;&nbsp;-&nbsp;array&nbsp;of&nbsp;input&nbsp;data<br>
 
67
&nbsp;&nbsp;&nbsp;&nbsp;t&nbsp;&nbsp;&nbsp;-&nbsp;array&nbsp;of&nbsp;targets<br>
 
68
Returns:<br>
 
69
&nbsp;&nbsp;&nbsp;&nbsp;sum-squared-error&nbsp;over&nbsp;all&nbsp;data</tt></dd></dl>
 
70
 
 
71
<dl><dt><a name="srn-train"><strong>train</strong></a>(self, x, t)</dt><dd><tt>Train&nbsp;a&nbsp;multilayer&nbsp;perceptron&nbsp;using&nbsp;scipy's&nbsp;leastsq&nbsp;optimizer<br>
 
72
Input:<br>
 
73
&nbsp;&nbsp;&nbsp;&nbsp;x&nbsp;&nbsp;&nbsp;-&nbsp;matrix&nbsp;of&nbsp;input&nbsp;data<br>
 
74
&nbsp;&nbsp;&nbsp;&nbsp;t&nbsp;&nbsp;&nbsp;-&nbsp;matrix&nbsp;of&nbsp;target&nbsp;outputs<br>
 
75
Returns:<br>
 
76
&nbsp;&nbsp;&nbsp;&nbsp;post-optimization&nbsp;weight&nbsp;vector</tt></dd></dl>
 
77
 
 
78
<dl><dt><a name="srn-unpack"><strong>unpack</strong></a>(self)</dt><dd><tt>Decompose&nbsp;1-d&nbsp;vector&nbsp;of&nbsp;weights&nbsp;w&nbsp;into&nbsp;appropriate&nbsp;weight&nbsp;<br>
 
79
matrices&nbsp;(w1,b1,w2,b2)&nbsp;and&nbsp;reinsert&nbsp;them&nbsp;into&nbsp;net</tt></dd></dl>
 
80
 
 
81
</td></tr></table></td></tr></table><p>
 
82
<table width="100%" cellspacing=0 cellpadding=2 border=0 summary="section">
 
83
<tr bgcolor="#eeaa77">
 
84
<td colspan=3 valign=bottom>&nbsp;<br>
 
85
<font color="#ffffff" face="helvetica, arial"><big><strong>Functions</strong></big></font></td></tr>
 
86
    
 
87
<tr><td bgcolor="#eeaa77"><tt>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</tt></td><td>&nbsp;</td>
 
88
<td width="100%"><dl><dt><a name="-main"><strong>main</strong></a>()</dt><dd><tt>Set&nbsp;up&nbsp;a&nbsp;1-2-1&nbsp;SRN&nbsp;to&nbsp;solve&nbsp;the&nbsp;temporal-XOR&nbsp;problem&nbsp;from&nbsp;Elman&nbsp;1990.</tt></dd></dl>
 
89
</td></tr></table>
 
90
</body></html>
 
 
b'\\ No newline at end of file'