forked from lucastheis/deepbelief
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdeepbelief-pysrc.html
183 lines (173 loc) · 16 KB
/
deepbelief-pysrc.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
<?xml version="1.0" encoding="ascii"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<title>deepbelief</title>
<link rel="stylesheet" href="epydoc.css" type="text/css" />
<script type="text/javascript" src="epydoc.js"></script>
</head>
<body bgcolor="white" text="black" link="blue" vlink="#204080"
alink="#204080">
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Home link -->
<th bgcolor="#70b0f0" class="navbar-select"
> Home </th>
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<!-- Project homepage -->
<th class="navbar" align="right" width="100%">
<table border="0" cellpadding="0" cellspacing="0">
<tr><th class="navbar" align="center"
>Deep Belief Net Toolbox</th>
</tr></table></th>
</tr>
</table>
<table width="100%" cellpadding="0" cellspacing="0">
<tr valign="top">
<td width="100%">
<span class="breadcrumbs">
Package deepbelief
</span>
</td>
<td>
<table cellpadding="0" cellspacing="0">
<!-- hide/show private -->
<tr><td align="right"><span class="options"
>[<a href="frames.html" target="_top">frames</a
>] | <a href="deepbelief-pysrc.html"
target="_top">no frames</a>]</span></td></tr>
</table>
</td>
</tr>
</table>
<h1 class="epydoc">Source Code for <a href="deepbelief-module.html">Package deepbelief</a></h1>
<pre class="py-src">
<a name="L1"></a><tt class="py-lineno"> 1</tt> <tt class="py-line"><tt class="py-docstring">"""</tt> </tt>
<a name="L2"></a><tt class="py-lineno"> 2</tt> <tt class="py-line"><tt class="py-docstring">Introduction</tt> </tt>
<a name="L3"></a><tt class="py-lineno"> 3</tt> <tt class="py-line"><tt class="py-docstring">============</tt> </tt>
<a name="L4"></a><tt class="py-lineno"> 4</tt> <tt class="py-line"><tt class="py-docstring"> This package implements algorithms for training and evaluating deep belief</tt> </tt>
<a name="L5"></a><tt class="py-lineno"> 5</tt> <tt class="py-line"><tt class="py-docstring"> networks (DBNs). Currently, the following variants of the restricted Boltzmann</tt> </tt>
<a name="L6"></a><tt class="py-lineno"> 6</tt> <tt class="py-line"><tt class="py-docstring"> machine are available for constructing DBNs:</tt> </tt>
<a name="L7"></a><tt class="py-lineno"> 7</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L8"></a><tt class="py-lineno"> 8</tt> <tt class="py-line"><tt class="py-docstring"> - L{RBM}</tt> </tt>
<a name="L9"></a><tt class="py-lineno"> 9</tt> <tt class="py-line"><tt class="py-docstring"> - L{GaussianRBM}</tt> </tt>
<a name="L10"></a><tt class="py-lineno">10</tt> <tt class="py-line"><tt class="py-docstring"> - L{SemiRBM}</tt> </tt>
<a name="L11"></a><tt class="py-lineno">11</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L12"></a><tt class="py-lineno">12</tt> <tt class="py-line"><tt class="py-docstring"> Also have a look a L{AbstractBM} which specifies much of the interface and</tt> </tt>
<a name="L13"></a><tt class="py-lineno">13</tt> <tt class="py-line"><tt class="py-docstring"> learning algorithms. In order to evaluate a trained DBN, the L{Estimator}</tt> </tt>
<a name="L14"></a><tt class="py-lineno">14</tt> <tt class="py-line"><tt class="py-docstring"> class can be used to estimate the the likelihood of the trained model.</tt> </tt>
<a name="L15"></a><tt class="py-lineno">15</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L16"></a><tt class="py-lineno">16</tt> <tt class="py-line"><tt class="py-docstring">Miscellaneous</tt> </tt>
<a name="L17"></a><tt class="py-lineno">17</tt> <tt class="py-line"><tt class="py-docstring">=============</tt> </tt>
<a name="L18"></a><tt class="py-lineno">18</tt> <tt class="py-line"><tt class="py-docstring"> For questions, comments or bug reports, contact U{Lucas Theis<mailto:[email protected]>}.</tt> </tt>
<a name="L19"></a><tt class="py-lineno">19</tt> <tt class="py-line"><tt class="py-docstring"> This code is published under the U{MIT License<http://www.opensource.org/licenses/mit-license.php>}.</tt> </tt>
<a name="L20"></a><tt class="py-lineno">20</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L21"></a><tt class="py-lineno">21</tt> <tt class="py-line"><tt class="py-docstring">Quick Tutorial</tt> </tt>
<a name="L22"></a><tt class="py-lineno">22</tt> <tt class="py-line"><tt class="py-docstring">==============</tt> </tt>
<a name="L23"></a><tt class="py-lineno">23</tt> <tt class="py-line"><tt class="py-docstring"> Assume that C{data} is a 10 by 1000 numpy matrix with real entries containing</tt> </tt>
<a name="L24"></a><tt class="py-lineno">24</tt> <tt class="py-line"><tt class="py-docstring"> 1000 data points. We want a deep belief network to learn the distribution of</tt> </tt>
<a name="L25"></a><tt class="py-lineno">25</tt> <tt class="py-line"><tt class="py-docstring"> the data.</tt> </tt>
<a name="L26"></a><tt class="py-lineno">26</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L27"></a><tt class="py-lineno">27</tt> <tt class="py-line"><tt class="py-docstring"> >>> print data.shape</tt> </tt>
<a name="L28"></a><tt class="py-lineno">28</tt> <tt class="py-line"><tt class="py-docstring"> (10, 1000)</tt> </tt>
<a name="L29"></a><tt class="py-lineno">29</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L30"></a><tt class="py-lineno">30</tt> <tt class="py-line"><tt class="py-docstring"> Import the relevant building blocks</tt> </tt>
<a name="L31"></a><tt class="py-lineno">31</tt> <tt class="py-line"><tt class="py-docstring"> for constructing deep belief networks.</tt> </tt>
<a name="L32"></a><tt class="py-lineno">32</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L33"></a><tt class="py-lineno">33</tt> <tt class="py-line"><tt class="py-docstring"> >>> from deepbelief import RBM, GaussianRBM, DBN</tt> </tt>
<a name="L34"></a><tt class="py-lineno">34</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L35"></a><tt class="py-lineno">35</tt> <tt class="py-line"><tt class="py-docstring"> Create a first layer with 10 visible units and 50 hidden units. The</tt> </tt>
<a name="L36"></a><tt class="py-lineno">36</tt> <tt class="py-line"><tt class="py-docstring"> L{GaussianRBM} is suited for modeling continuous data.</tt> </tt>
<a name="L37"></a><tt class="py-lineno">37</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L38"></a><tt class="py-lineno">38</tt> <tt class="py-line"><tt class="py-docstring"> >>> dbn = DBN(GaussianRBM(10, 50))</tt> </tt>
<a name="L39"></a><tt class="py-lineno">39</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L40"></a><tt class="py-lineno">40</tt> <tt class="py-line"><tt class="py-docstring"> Split the data into batches of size 10 and train the first layer for 50</tt> </tt>
<a name="L41"></a><tt class="py-lineno">41</tt> <tt class="py-line"><tt class="py-docstring"> iterations.</tt> </tt>
<a name="L42"></a><tt class="py-lineno">42</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L43"></a><tt class="py-lineno">43</tt> <tt class="py-line"><tt class="py-docstring"> >>> dbn.train(data, num_epochs=50, batch_size=10)</tt> </tt>
<a name="L44"></a><tt class="py-lineno">44</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L45"></a><tt class="py-lineno">45</tt> <tt class="py-line"><tt class="py-docstring"> Add a second layer to the network.</tt> </tt>
<a name="L46"></a><tt class="py-lineno">46</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L47"></a><tt class="py-lineno">47</tt> <tt class="py-line"><tt class="py-docstring"> >>> dbn.add_layer(RBM(50, 50))</tt> </tt>
<a name="L48"></a><tt class="py-lineno">48</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L49"></a><tt class="py-lineno">49</tt> <tt class="py-line"><tt class="py-docstring"> Train the second layer.</tt> </tt>
<a name="L50"></a><tt class="py-lineno">50</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L51"></a><tt class="py-lineno">51</tt> <tt class="py-line"><tt class="py-docstring"> >>> dbn.train(data, num_epochs=50, batch_size=10)</tt> </tt>
<a name="L52"></a><tt class="py-lineno">52</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L53"></a><tt class="py-lineno">53</tt> <tt class="py-line"><tt class="py-docstring"> Generate another 500 data points by sampling from the trained model.</tt> </tt>
<a name="L54"></a><tt class="py-lineno">54</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L55"></a><tt class="py-lineno">55</tt> <tt class="py-line"><tt class="py-docstring"> >>> samples = dbn.sample(500)</tt> </tt>
<a name="L56"></a><tt class="py-lineno">56</tt> <tt class="py-line"><tt class="py-docstring">"""</tt> </tt>
<a name="L57"></a><tt class="py-lineno">57</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-0" class="py-name" targets="Module deepbelief.dbn=deepbelief.dbn-module.html"><a title="deepbelief.dbn" class="py-name" href="#" onclick="return doclink('link-0', 'dbn', 'link-0');">dbn</a></tt> <tt class="py-keyword">import</tt> <tt id="link-1" class="py-name" targets="Class deepbelief.dbn.DBN=deepbelief.dbn.DBN-class.html"><a title="deepbelief.dbn.DBN" class="py-name" href="#" onclick="return doclink('link-1', 'DBN', 'link-1');">DBN</a></tt> </tt>
<a name="L58"></a><tt class="py-lineno">58</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-2" class="py-name" targets="Module deepbelief.rbm=deepbelief.rbm-module.html"><a title="deepbelief.rbm" class="py-name" href="#" onclick="return doclink('link-2', 'rbm', 'link-2');">rbm</a></tt> <tt class="py-keyword">import</tt> <tt id="link-3" class="py-name" targets="Class deepbelief.rbm.RBM=deepbelief.rbm.RBM-class.html"><a title="deepbelief.rbm.RBM" class="py-name" href="#" onclick="return doclink('link-3', 'RBM', 'link-3');">RBM</a></tt> </tt>
<a name="L59"></a><tt class="py-lineno">59</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-4" class="py-name" targets="Module deepbelief.gaussianrbm=deepbelief.gaussianrbm-module.html"><a title="deepbelief.gaussianrbm" class="py-name" href="#" onclick="return doclink('link-4', 'gaussianrbm', 'link-4');">gaussianrbm</a></tt> <tt class="py-keyword">import</tt> <tt id="link-5" class="py-name" targets="Class deepbelief.gaussianrbm.GaussianRBM=deepbelief.gaussianrbm.GaussianRBM-class.html"><a title="deepbelief.gaussianrbm.GaussianRBM" class="py-name" href="#" onclick="return doclink('link-5', 'GaussianRBM', 'link-5');">GaussianRBM</a></tt> </tt>
<a name="L60"></a><tt class="py-lineno">60</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-6" class="py-name" targets="Module deepbelief.semirbm=deepbelief.semirbm-module.html"><a title="deepbelief.semirbm" class="py-name" href="#" onclick="return doclink('link-6', 'semirbm', 'link-6');">semirbm</a></tt> <tt class="py-keyword">import</tt> <tt id="link-7" class="py-name" targets="Class deepbelief.semirbm.SemiRBM=deepbelief.semirbm.SemiRBM-class.html"><a title="deepbelief.semirbm.SemiRBM" class="py-name" href="#" onclick="return doclink('link-7', 'SemiRBM', 'link-7');">SemiRBM</a></tt> </tt>
<a name="L61"></a><tt class="py-lineno">61</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-8" class="py-name" targets="Module deepbelief.estimator=deepbelief.estimator-module.html"><a title="deepbelief.estimator" class="py-name" href="#" onclick="return doclink('link-8', 'estimator', 'link-8');">estimator</a></tt> <tt class="py-keyword">import</tt> <tt id="link-9" class="py-name" targets="Class deepbelief.estimator.Estimator=deepbelief.estimator.Estimator-class.html"><a title="deepbelief.estimator.Estimator" class="py-name" href="#" onclick="return doclink('link-9', 'Estimator', 'link-9');">Estimator</a></tt> </tt>
<a name="L62"></a><tt class="py-lineno">62</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-10" class="py-name" targets="Module deepbelief.mixbm=deepbelief.mixbm-module.html"><a title="deepbelief.mixbm" class="py-name" href="#" onclick="return doclink('link-10', 'mixbm', 'link-10');">mixbm</a></tt> <tt class="py-keyword">import</tt> <tt id="link-11" class="py-name" targets="Class deepbelief.mixbm.MixBM=deepbelief.mixbm.MixBM-class.html"><a title="deepbelief.mixbm.MixBM" class="py-name" href="#" onclick="return doclink('link-11', 'MixBM', 'link-11');">MixBM</a></tt> </tt>
<a name="L63"></a><tt class="py-lineno">63</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-12" class="py-name" targets="Module deepbelief.abstractbm=deepbelief.abstractbm-module.html"><a title="deepbelief.abstractbm" class="py-name" href="#" onclick="return doclink('link-12', 'abstractbm', 'link-12');">abstractbm</a></tt> <tt class="py-keyword">import</tt> <tt id="link-13" class="py-name" targets="Class deepbelief.abstractbm.AbstractBM=deepbelief.abstractbm.AbstractBM-class.html"><a title="deepbelief.abstractbm.AbstractBM" class="py-name" href="#" onclick="return doclink('link-13', 'AbstractBM', 'link-13');">AbstractBM</a></tt> </tt>
<a name="L64"></a><tt class="py-lineno">64</tt> <tt class="py-line"> </tt>
<a name="L65"></a><tt class="py-lineno">65</tt> <tt class="py-line"><tt class="py-name">__docformat__</tt> <tt class="py-op">=</tt> <tt class="py-string">'epytext'</tt> </tt>
<a name="L66"></a><tt class="py-lineno">66</tt> <tt class="py-line"> </tt><script type="text/javascript">
<!--
expandto(location.href);
// -->
</script>
</pre>
<br />
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Home link -->
<th bgcolor="#70b0f0" class="navbar-select"
> Home </th>
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<!-- Project homepage -->
<th class="navbar" align="right" width="100%">
<table border="0" cellpadding="0" cellspacing="0">
<tr><th class="navbar" align="center"
>Deep Belief Net Toolbox</th>
</tr></table></th>
</tr>
</table>
<table border="0" cellpadding="0" cellspacing="0" width="100%%">
<tr>
<td align="left" class="footer">
Generated by Epydoc 3.0.1 on Thu Jun 9 17:26:46 2011
</td>
<td align="right" class="footer">
<a target="mainFrame" href="http://epydoc.sourceforge.net"
>http://epydoc.sourceforge.net</a>
</td>
</tr>
</table>
<script type="text/javascript">
<!--
// Private objects are initially displayed (because if
// javascript is turned off then we want them to be
// visible); but by default, we want to hide them. So hide
// them unless we have a cookie that says to show them.
checkCookie();
// -->
</script>
</body>
</html>