mirror of
https://github.com/SheffieldML/GPy.git
synced 2026-05-05 17:52:39 +02:00
update kernel tutorial
This commit is contained in:
Homer Strong
parent
3a150198e8
commit
1d7712ecc8
372 changed files with 92313 additions and 121 deletions
453
doc/_build/html/tuto_interacting_with_models.html
vendored
Normal file
453
doc/_build/html/tuto_interacting_with_models.html
vendored
Normal file
|
|
@ -0,0 +1,453 @@
|
|||
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
|
||||
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
|
||||
|
||||
|
||||
<html xmlns="http://www.w3.org/1999/xhtml">
|
||||
<head>
|
||||
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
|
||||
|
||||
<title>Interacting with models — GPy documentation</title>
|
||||
|
||||
<link rel="stylesheet" href="_static//default.css" type="text/css" />
|
||||
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
|
||||
|
||||
<script type="text/javascript">
|
||||
var DOCUMENTATION_OPTIONS = {
|
||||
URL_ROOT: './',
|
||||
VERSION: '',
|
||||
COLLAPSE_INDEX: false,
|
||||
FILE_SUFFIX: '.html',
|
||||
HAS_SOURCE: true
|
||||
};
|
||||
</script>
|
||||
<script type="text/javascript" src="_static/jquery.js"></script>
|
||||
<script type="text/javascript" src="_static/underscore.js"></script>
|
||||
<script type="text/javascript" src="_static/doctools.js"></script>
|
||||
<link rel="top" title="GPy documentation" href="index.html" />
|
||||
</head>
|
||||
<body role="document">
|
||||
<div class="related" role="navigation" aria-label="related navigation">
|
||||
<h3>Navigation</h3>
|
||||
<ul>
|
||||
<li class="right" style="margin-right: 10px">
|
||||
<a href="genindex.html" title="General Index"
|
||||
accesskey="I">index</a></li>
|
||||
<li class="right" >
|
||||
<a href="py-modindex.html" title="Python Module Index"
|
||||
>modules</a> |</li>
|
||||
<li class="nav-item nav-item-0"><a href="index.html">GPy documentation</a> »</li>
|
||||
</ul>
|
||||
</div>
|
||||
|
||||
<div class="document">
|
||||
<div class="documentwrapper">
|
||||
<div class="bodywrapper">
|
||||
<div class="body" role="main">
|
||||
|
||||
<div class="section" id="interacting-with-models">
|
||||
<span id="id1"></span><h1>Interacting with models<a class="headerlink" href="#interacting-with-models" title="Permalink to this headline">¶</a></h1>
|
||||
<p>The GPy model class has a set of features which are
|
||||
designed to make it simple to explore the parameter
|
||||
space of the model. By default, the scipy optimisers
|
||||
are used to fit GPy models (via model.optimize()),
|
||||
for which we provide mechanisms for ‘free’ optimisation:
|
||||
GPy can ensure that naturally positive parameters
|
||||
(such as variances) remain positive. But these mechanisms
|
||||
are much more powerful than simple reparameterisation,
|
||||
as we shall see.</p>
|
||||
<p>Along this tutorial we’ll use a sparse GP regression model
|
||||
as example. This example can be in <code class="docutils literal"><span class="pre">GPy.examples.regression</span></code>.
|
||||
All of the examples included in GPy return an instance
|
||||
of a model class, and therefore they can be called in
|
||||
the following way:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="kn">import</span> <span class="nn">numpy</span> <span class="kn">as</span> <span class="nn">np</span>
|
||||
<span class="kn">import</span> <span class="nn">pylab</span> <span class="kn">as</span> <span class="nn">pb</span>
|
||||
<span class="n">pb</span><span class="o">.</span><span class="n">ion</span><span class="p">()</span>
|
||||
<span class="kn">import</span> <span class="nn">GPy</span>
|
||||
<span class="n">m</span> <span class="o">=</span> <span class="n">GPy</span><span class="o">.</span><span class="n">examples</span><span class="o">.</span><span class="n">regression</span><span class="o">.</span><span class="n">sparse_GP_regression_1D</span><span class="p">()</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<div class="section" id="examining-the-model-using-print">
|
||||
<h2>Examining the model using print<a class="headerlink" href="#examining-the-model-using-print" title="Permalink to this headline">¶</a></h2>
|
||||
<p>To see the current state of the model parameters,
|
||||
and the model’s (marginal) likelihood just print the model</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="k">print</span> <span class="n">m</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>The first thing displayed on the screen is the log-likelihood
|
||||
value of the model with its current parameters. Below the
|
||||
log-likelihood, a table with all the model’s parameters
|
||||
is shown. For each parameter, the table contains the name
|
||||
of the parameter, the current value, and in case there are
|
||||
defined: constraints, ties and prior distrbutions associated.</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre>Name : sparse gp
|
||||
Log-likelihood : 588.947189413
|
||||
Number of Parameters : 8
|
||||
Parameters:
|
||||
sparse_gp. | Value | Constraint | Prior | Tied to
|
||||
inducing inputs | (5, 1) | | |
|
||||
rbf.variance | 1.91644016819 | +ve | |
|
||||
rbf.lengthscale | 2.62103621347 | +ve | |
|
||||
Gaussian_noise.variance | 0.00269870373421 | +ve | |
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>In this case the kernel parameters (<code class="docutils literal"><span class="pre">rbf.variance</span></code>,
|
||||
<code class="docutils literal"><span class="pre">rbf.lengthscale</span></code>) as well as
|
||||
the likelihood noise parameter (<code class="docutils literal"><span class="pre">Gaussian_noise.variance</span></code>), are constrained
|
||||
to be positive, while the inducing inputs have no
|
||||
constraints associated. Also there are no ties or prior defined.</p>
|
||||
<p>You can also print all subparts of the model, by printing the
|
||||
subcomponents individually:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="k">print</span> <span class="n">m</span><span class="o">.</span><span class="n">rbf</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>This will print the details of this particular parameter handle:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre>rbf. | Value | Constraint | Prior | Tied to
|
||||
variance | 1.91644016819 | +ve | |
|
||||
lengthscale | 2.62103621347 | +ve | |
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>When you want to get a closer look into
|
||||
multivalue parameters, print them directly:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre>print m.inducing_inputs
|
||||
|
||||
Index | sparse_gp.inducing_inputs | Constraint | Prior | Tied to
|
||||
[0 0] | 2.7189499 | | | N/A
|
||||
[1 0] | 0.02006533 | | | N/A
|
||||
[2 0] | -1.5299386 | | | N/A
|
||||
[3 0] | -2.7001675 | | | N/A
|
||||
[4 0] | 1.4654162 | | | N/A
|
||||
</pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="section" id="interacting-with-parameters">
|
||||
<h2>Interacting with Parameters:<a class="headerlink" href="#interacting-with-parameters" title="Permalink to this headline">¶</a></h2>
|
||||
<p>The preferred way of interacting with parameters is to act on the
|
||||
parameter handle itself.
|
||||
Interacting with parameter handles is simple. The names, printed by <cite>print m</cite>
|
||||
are accessible interactively and programatically. For example try to
|
||||
set kernels (<cite>rbf</cite>) <cite>lengthscale</cite> to <cite>.2</cite> and print the result:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="n">m</span><span class="o">.</span><span class="n">rbf</span><span class="o">.</span><span class="n">lengthscale</span> <span class="o">=</span> <span class="o">.</span><span class="mi">2</span>
|
||||
<span class="k">print</span> <span class="n">m</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>You should see this:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre>Name : sparse gp
|
||||
Log-likelihood : 588.947189413
|
||||
Number of Parameters : 8
|
||||
Parameters:
|
||||
sparse_gp. | Value | Constraint | Prior | Tied to
|
||||
inducing inputs | (5, 1) | | |
|
||||
rbf.variance | 1.91644016819 | +ve | |
|
||||
rbf.lengthscale | 0.2 | +ve | |
|
||||
Gaussian_noise.variance | 0.00269870373421 | +ve | |
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>This will already have updated the model’s inner state, so you can
|
||||
plot it or see the changes in the posterior <cite>m.posterior</cite> of the model.</p>
|
||||
<div class="section" id="regular-expressions">
|
||||
<h3>Regular expressions<a class="headerlink" href="#regular-expressions" title="Permalink to this headline">¶</a></h3>
|
||||
<p>The model’s parameters can also be accessed through regular
|
||||
expressions, by ‘indexing’ the model with a regular expression,
|
||||
matching the parameter name. Through indexing by regular expression,
|
||||
you can only retrieve leafs of the hierarchy, and you can retrieve the
|
||||
values matched by calling <cite>values()</cite> on the returned object:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span>
|
||||
<span class="go"> Index | sparse_gp.rbf.variance | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> [0] | 2.1500132 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.Gaussian_noise.variance | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 0.0024268215 | | | N/A</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span><span class="o">.</span><span class="n">values</span><span class="p">()</span>
|
||||
<span class="go">[ 2.1500132 0.00242682]</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="p">[</span><span class="s">'rbf'</span><span class="p">]</span>
|
||||
<span class="go"> Index | sparse_gp.rbf.variance | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> [0] | 2.1500132 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.rbf.lengthscale | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 2.6782803 | | | N/A</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>There is access to setting parameters by regular expression,
|
||||
as well. Here are a few examples of how to set parameters by regular expression:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span> <span class="o">=</span> <span class="o">.</span><span class="mi">1</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span>
|
||||
<span class="go"> Index | sparse_gp.rbf.variance | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> [0] | 0.1 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.Gaussian_noise.variance | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 0.1 | | | N/A</span>
|
||||
<span class="gp">>>> </span><span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span> <span class="o">=</span> <span class="p">[</span><span class="o">.</span><span class="mi">1</span><span class="p">,</span> <span class="o">.</span><span class="mi">2</span><span class="p">]</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span>
|
||||
<span class="go"> Index | sparse_gp.rbf.variance | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> [0] | 0.1 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.Gaussian_noise.variance | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 0.2 | | | N/A</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>The fact that only leaf nodes can be accesses we can print all
|
||||
parameters in a flattened view, by printing the regular expression
|
||||
match of matching all objects:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="p">[</span><span class="s">''</span><span class="p">]</span>
|
||||
<span class="go"> Index | sparse_gp.inducing_inputs | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> [0 0] | -2.6716041 | | | N/A</span>
|
||||
<span class="go"> [1 0] | -1.4665111 | | | N/A</span>
|
||||
<span class="go"> [2 0] | -0.031010293 | | | N/A</span>
|
||||
<span class="go"> [3 0] | 1.4563711 | | | N/A</span>
|
||||
<span class="go"> [4 0] | 2.6803046 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.rbf.variance | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 0.1 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.rbf.lengthscale | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 2.6782803 | | | N/A</span>
|
||||
<span class="go"> ----- | sparse_gp.Gaussian_noise.variance | ---------- | ---------- | -------</span>
|
||||
<span class="go"> [0] | 0.2 | | | N/A</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="section" id="setting-and-fetching-parameters-parameter-array">
|
||||
<h3>Setting and fetching parameters <cite>parameter_array</cite><a class="headerlink" href="#setting-and-fetching-parameters-parameter-array" title="Permalink to this headline">¶</a></h3>
|
||||
<p>Another way to interact with the model’s parameters is through the
|
||||
<cite>parameter_array</cite>. The Parameter array holds all the parameters of the
|
||||
model in one place and is editable. It can be accessed through
|
||||
indexing the model for example you can set all the parameters through
|
||||
this mechanism:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">new_params</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">r_</span><span class="p">[[</span><span class="o">-</span><span class="mi">4</span><span class="p">,</span><span class="o">-</span><span class="mi">2</span><span class="p">,</span><span class="mi">0</span><span class="p">,</span><span class="mi">2</span><span class="p">,</span><span class="mi">4</span><span class="p">],</span> <span class="p">[</span><span class="o">.</span><span class="mi">5</span><span class="p">,</span><span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="o">.</span><span class="mi">3</span><span class="p">]]</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">new_params</span>
|
||||
<span class="go">array([-4. , -2. , 0. , 2. , 4. , 0.5, 2. , 0.3])</span>
|
||||
<span class="gp">>>> </span><span class="n">m</span><span class="p">[:]</span> <span class="o">=</span> <span class="n">new_params</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span>
|
||||
<span class="go">Name : sparse gp</span>
|
||||
<span class="go">Log-likelihood : -147.561160209</span>
|
||||
<span class="go">Number of Parameters : 8</span>
|
||||
<span class="go">Parameters:</span>
|
||||
<span class="go"> sparse_gp. | Value | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> inducing inputs | (5, 1) | | |</span>
|
||||
<span class="go"> rbf.variance | 0.5 | +sq | |</span>
|
||||
<span class="go"> rbf.lengthscale | 2.0 | +ve | |</span>
|
||||
<span class="go"> Gaussian_noise.variance | 0.3 | +sq | |</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>Parameters themselves (leafs of the hierarchy) can be indexed and used
|
||||
the same way as numpy arrays. First let us set a slice of the
|
||||
<cite>inducing_inputs</cite>:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">inducing_inputs</span><span class="p">[</span><span class="mi">2</span><span class="p">:,</span> <span class="mi">0</span><span class="p">]</span> <span class="o">=</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span><span class="mi">3</span><span class="p">,</span><span class="mi">5</span><span class="p">]</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="o">.</span><span class="n">inducing_indputs</span>
|
||||
<span class="go"> Index | sparse_gp.inducing_inputs | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> [0 0] | -4 | | | N/A</span>
|
||||
<span class="go"> [1 0] | -2 | | | N/A</span>
|
||||
<span class="go"> [2 0] | 1 | | | N/A</span>
|
||||
<span class="go"> [3 0] | 3 | | | N/A</span>
|
||||
<span class="go"> [4 0] | 5 | | | N/A</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>Or you use the parameters as normal numpy arrays for calculations:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">precision</span> <span class="o">=</span> <span class="mf">1.</span><span class="o">/</span><span class="n">m</span><span class="o">.</span><span class="n">Gaussian_noise</span><span class="o">.</span><span class="n">variance</span>
|
||||
<span class="go">array([ 3.33333333])</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="section" id="getting-the-model-s-log-likelihood">
|
||||
<h2>Getting the model’s log likelihood<a class="headerlink" href="#getting-the-model-s-log-likelihood" title="Permalink to this headline">¶</a></h2>
|
||||
<p>Appart form the printing the model, the marginal
|
||||
log-likelihood can be obtained by using the function
|
||||
<code class="docutils literal"><span class="pre">log_likelihood()</span></code>.:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">log_likelihood</span><span class="p">()</span>
|
||||
<span class="go">array([-152.83377316])</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>If you want to ensure the log likelihood as a float, call <cite>float()</cite>
|
||||
around it:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="nb">float</span><span class="p">(</span><span class="n">m</span><span class="o">.</span><span class="n">log_likelihood</span><span class="p">())</span>
|
||||
<span class="go">-152.83377316356177</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="section" id="getting-the-model-parameter-s-gradients">
|
||||
<h2>Getting the model parameter’s gradients<a class="headerlink" href="#getting-the-model-parameter-s-gradients" title="Permalink to this headline">¶</a></h2>
|
||||
<p>The gradients of a model can shed light on understanding the
|
||||
(possibly hard) optimization process. The gradients of each parameter
|
||||
handle can be accessed through their <cite>gradient</cite> field.:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="o">.</span><span class="n">gradient</span>
|
||||
<span class="go">[ 5.51170031 9.71735112 -4.20282106 -3.45667035 -1.58828165</span>
|
||||
<span class="go"> -2.11549358 12.40292787 -627.75467803]</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="o">.</span><span class="n">rbf</span><span class="o">.</span><span class="n">gradient</span>
|
||||
<span class="go">[ -2.11549358 12.40292787]</span>
|
||||
<span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">optimize</span><span class="p">()</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span><span class="o">.</span><span class="n">gradient</span>
|
||||
<span class="go">[ -5.98046560e-04 -3.64576085e-04 1.98005930e-04 3.43381219e-04</span>
|
||||
<span class="go">-6.85685104e-04 -1.28800748e-05 1.08552429e-03 2.74058081e-01]</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="section" id="adjusting-the-model-s-constraints">
|
||||
<h2>Adjusting the model’s constraints<a class="headerlink" href="#adjusting-the-model-s-constraints" title="Permalink to this headline">¶</a></h2>
|
||||
<p>When we initially call the example, it was optimized and hence the
|
||||
log-likelihood gradients were close to zero. However, since
|
||||
we have been changing the parameters, the gradients are far from zero now.
|
||||
Next we are going to show how to optimize the model setting different
|
||||
restrictions on the parameters.</p>
|
||||
<p>Once a constraint has been set on a parameter, it is possible to remove
|
||||
it with the command <code class="docutils literal"><span class="pre">unconstrain()</span></code>, which can be called on any
|
||||
parameter handle of the model. The methods <cite>constrain()</cite> and
|
||||
<cite>unconstrain()</cite> return the indices which were actually unconstrained,
|
||||
relative to the parameter handle the method was called on. This is
|
||||
particularly handy for reporting which parameters where reconstrained,
|
||||
when reconstraining a parameter, which was already constrained:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">rbf</span><span class="o">.</span><span class="n">variance</span><span class="o">.</span><span class="n">unconstrain</span><span class="p">()</span>
|
||||
<span class="go">array([0])</span>
|
||||
<span class="go">>>>m.unconstrain()</span>
|
||||
<span class="go">array([6, 7])</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>If you want to unconstrain only a specific constraint, you can pass it
|
||||
as an argument of <code class="docutils literal"><span class="pre">unconstrain(Transformation)</span></code> (<code class="xref py py-class docutils literal"><span class="pre">Transformation</span></code>), or call
|
||||
the respective method, such as <code class="docutils literal"><span class="pre">unconstrain_fixed()</span></code> (or
|
||||
<code class="docutils literal"><span class="pre">unfix()</span></code>) to only unfix fixed parameters.:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">inducing_input</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">fix</span><span class="p">()</span>
|
||||
<span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">unfix</span><span class="p">()</span>
|
||||
<span class="gp">>>> </span><span class="n">m</span><span class="o">.</span><span class="n">rbf</span><span class="o">.</span><span class="n">constrain_positive</span><span class="p">()</span>
|
||||
<span class="gp">>>> </span><span class="k">print</span> <span class="n">m</span>
|
||||
<span class="go">Name : sparse gp</span>
|
||||
<span class="go">Log-likelihood : 620.741066698</span>
|
||||
<span class="go">Number of Parameters : 8</span>
|
||||
<span class="go">Parameters:</span>
|
||||
<span class="go"> sparse_gp. | Value | Constraint | Prior | Tied to</span>
|
||||
<span class="go"> inducing inputs | (5, 1) | | |</span>
|
||||
<span class="go"> rbf.variance | 1.48329711218 | +ve | |</span>
|
||||
<span class="go"> rbf.lengthscale | 2.5430947048 | +ve | |</span>
|
||||
<span class="go"> Gaussian_noise.variance | 0.00229714444128 | | |</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>As you can see, <code class="docutils literal"><span class="pre">unfix()</span></code> only unfixed the inducing_input, and did
|
||||
not change the positive constraint of the kernel.</p>
|
||||
<p>The parameter handles come with default constraints, so you will
|
||||
rarely be needing to adjust the constraints of a model. In the rare
|
||||
cases of needing to adjust the constraints of a model, or in need of
|
||||
fixing some parameters, you can do so with the functions
|
||||
<code class="docutils literal"><span class="pre">constrain_{positive|negative|bounded|fixed}()</span></code>.:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="n">m</span><span class="p">[</span><span class="s">'.*var'</span><span class="p">]</span><span class="o">.</span><span class="n">constrain_positive</span><span class="p">()</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="section" id="available-constraints">
|
||||
<h2>Available Constraints<a class="headerlink" href="#available-constraints" title="Permalink to this headline">¶</a></h2>
|
||||
<ul class="simple">
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">Logexp()</span></code></li>
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">Exponent()</span></code></li>
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">Square()</span></code></li>
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">Logistic()</span></code></li>
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">LogexpNeg()</span></code></li>
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">NegativeExponent()</span></code></li>
|
||||
<li><code class="xref py py-meth docutils literal"><span class="pre">NegativeLogexp()</span></code></li>
|
||||
</ul>
|
||||
</div>
|
||||
<div class="section" id="tying-parameters">
|
||||
<h2>Tying Parameters<a class="headerlink" href="#tying-parameters" title="Permalink to this headline">¶</a></h2>
|
||||
<p>Not yet implemented for GPy version 0.6.0</p>
|
||||
</div>
|
||||
<div class="section" id="optimizing-the-model">
|
||||
<h2>Optimizing the model<a class="headerlink" href="#optimizing-the-model" title="Permalink to this headline">¶</a></h2>
|
||||
<p>Once we have finished defining the constraints,
|
||||
we can now optimize the model with the function
|
||||
<code class="docutils literal"><span class="pre">optimize</span></code>.:</p>
|
||||
<div class="highlight-python"><div class="highlight"><pre><span class="n">m</span><span class="o">.</span><span class="n">Gaussian_noise</span><span class="o">.</span><span class="n">constrain_positive</span><span class="p">()</span>
|
||||
<span class="n">m</span><span class="o">.</span><span class="n">rbf</span><span class="o">.</span><span class="n">constrain_positive</span><span class="p">()</span>
|
||||
<span class="n">m</span><span class="o">.</span><span class="n">optimize</span><span class="p">()</span>
|
||||
</pre></div>
|
||||
</div>
|
||||
<p>By deafult, GPy uses the lbfgsb optimizer.</p>
|
||||
<p>Some optional parameters may be discussed here.</p>
|
||||
<ul class="simple">
|
||||
<li><code class="docutils literal"><span class="pre">optimizer</span></code>: which optimizer to use, currently there are <code class="docutils literal"><span class="pre">lbfgsb,</span> <span class="pre">fmin_tnc,</span>
|
||||
<span class="pre">scg,</span> <span class="pre">simplex</span></code> or any unique identifier uniquely identifying an
|
||||
optimizer. Thus, you can say <code class="docutils literal"><span class="pre">m.optimize('bfgs')</span> <span class="pre">for</span> <span class="pre">using</span> <span class="pre">the</span>
|
||||
<span class="pre">``lbfgsb</span></code> optimizer</li>
|
||||
<li><code class="docutils literal"><span class="pre">messages</span></code>: if the optimizer is verbose. Each optimizer has its
|
||||
own way of printing, so do not be confused by differing messages of
|
||||
different optimizers</li>
|
||||
<li><code class="docutils literal"><span class="pre">max_iters</span></code>: Maximum number of iterations to take. Some optimizers
|
||||
see iterations as function calls, others as iterations of the
|
||||
algorithm. Please be advised to look into <code class="docutils literal"><span class="pre">scipy.optimize</span></code> for
|
||||
more instructions, if the number of iterations matter, so you can
|
||||
give the right parameters to <code class="docutils literal"><span class="pre">optimize()</span></code></li>
|
||||
<li><code class="docutils literal"><span class="pre">gtol</span></code>: only for some optimizers. Will determine the convergence
|
||||
criterion, as the tolerance of gradient to finish the optimization.</li>
|
||||
</ul>
|
||||
</div>
|
||||
<div class="section" id="further-reading">
|
||||
<h2>Further Reading<a class="headerlink" href="#further-reading" title="Permalink to this headline">¶</a></h2>
|
||||
<p>All of the mechansiams for dealing
|
||||
with parameters are baked right into GPy.core.model, from which all of
|
||||
the classes in GPy.models inherrit. To learn how to construct your own
|
||||
model, you might want to read <a class="reference internal" href="tuto_creating_new_models.html#creating-new-models"><span>Creating new Models</span></a>. If you want
|
||||
to learn how to create kernels, please refer to
|
||||
<span class="xref std std-ref">creating_new_kernels</span></p>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
<div class="sphinxsidebar" role="navigation" aria-label="main navigation">
|
||||
<div class="sphinxsidebarwrapper">
|
||||
<h3><a href="index.html">Table Of Contents</a></h3>
|
||||
<ul>
|
||||
<li><a class="reference internal" href="#">Interacting with models</a><ul>
|
||||
<li><a class="reference internal" href="#examining-the-model-using-print">Examining the model using print</a></li>
|
||||
<li><a class="reference internal" href="#interacting-with-parameters">Interacting with Parameters:</a><ul>
|
||||
<li><a class="reference internal" href="#regular-expressions">Regular expressions</a></li>
|
||||
<li><a class="reference internal" href="#setting-and-fetching-parameters-parameter-array">Setting and fetching parameters <cite>parameter_array</cite></a></li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><a class="reference internal" href="#getting-the-model-s-log-likelihood">Getting the model’s log likelihood</a></li>
|
||||
<li><a class="reference internal" href="#getting-the-model-parameter-s-gradients">Getting the model parameter’s gradients</a></li>
|
||||
<li><a class="reference internal" href="#adjusting-the-model-s-constraints">Adjusting the model’s constraints</a></li>
|
||||
<li><a class="reference internal" href="#available-constraints">Available Constraints</a></li>
|
||||
<li><a class="reference internal" href="#tying-parameters">Tying Parameters</a></li>
|
||||
<li><a class="reference internal" href="#optimizing-the-model">Optimizing the model</a></li>
|
||||
<li><a class="reference internal" href="#further-reading">Further Reading</a></li>
|
||||
</ul>
|
||||
</li>
|
||||
</ul>
|
||||
|
||||
<div role="note" aria-label="source link">
|
||||
<h3>This Page</h3>
|
||||
<ul class="this-page-menu">
|
||||
<li><a href="_sources/tuto_interacting_with_models.txt"
|
||||
rel="nofollow">Show Source</a></li>
|
||||
</ul>
|
||||
</div>
|
||||
<div id="searchbox" style="display: none" role="search">
|
||||
<h3>Quick search</h3>
|
||||
<form class="search" action="search.html" method="get">
|
||||
<input type="text" name="q" />
|
||||
<input type="submit" value="Go" />
|
||||
<input type="hidden" name="check_keywords" value="yes" />
|
||||
<input type="hidden" name="area" value="default" />
|
||||
</form>
|
||||
<p class="searchtip" style="font-size: 90%">
|
||||
Enter search terms or a module, class or function name.
|
||||
</p>
|
||||
</div>
|
||||
<script type="text/javascript">$('#searchbox').show(0);</script>
|
||||
</div>
|
||||
</div>
|
||||
<div class="clearer"></div>
|
||||
</div>
|
||||
<div class="related" role="navigation" aria-label="related navigation">
|
||||
<h3>Navigation</h3>
|
||||
<ul>
|
||||
<li class="right" style="margin-right: 10px">
|
||||
<a href="genindex.html" title="General Index"
|
||||
>index</a></li>
|
||||
<li class="right" >
|
||||
<a href="py-modindex.html" title="Python Module Index"
|
||||
>modules</a> |</li>
|
||||
<li class="nav-item nav-item-0"><a href="index.html">GPy documentation</a> »</li>
|
||||
</ul>
|
||||
</div>
|
||||
<div class="footer" role="contentinfo">
|
||||
© Copyright 2013, Author.
|
||||
Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.3.1.
|
||||
</div>
|
||||
</body>
|
||||
</html>
|
||||
Loading…
Add table
Add a link
Reference in a new issue