mirror of
https://github.com/fofolee/uTools-Manuals.git
synced 2025-06-18 22:06:57 +08:00
83 lines
3.3 KiB
HTML
83 lines
3.3 KiB
HTML
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
|
|
<html>
|
|
<head>
|
|
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
|
|
<title>使用 Widrow 和 Nguyen 算法初始化权重。</title>
|
|
</head>
|
|
<body class="docs"><div id="layout">
|
|
<div id="layout-content"><div id="function.fann-init-weights" class="refentry">
|
|
<div class="refnamediv">
|
|
<h1 class="refname">fann_init_weights</h1>
|
|
<p class="verinfo">(PECL fann >= 1.0.0)</p><p class="refpurpose"><span class="refname">fann_init_weights</span> — <span class="dc-title">使用 Widrow 和 Nguyen 算法初始化权重。</span></p>
|
|
|
|
</div>
|
|
|
|
<div class="refsect1 description" id="refsect1-function.fann-init-weights-description">
|
|
<h3 class="title">说明</h3>
|
|
<div class="methodsynopsis dc-description">
|
|
<span class="methodname"><strong>fann_init_weights</strong></span>
|
|
( <span class="methodparam"><span class="type">resource</span> <code class="parameter">$ann</code></span>
|
|
, <span class="methodparam"><span class="type">resource</span> <code class="parameter">$train_data</code></span>
|
|
) : <span class="type">bool</span></div>
|
|
|
|
<p class="para rdfs-comment">
|
|
使用 Widrow 和 Nguyen 算法初始化权重。
|
|
</p>
|
|
<p class="para">
|
|
该函数的作用和 <span class="function"><a href="fann_randomize_weights.html" class="function">fann_randomize_weights()</a></span>函数相似。 该函数将会使用
|
|
Derrick Nguyen 和 Bernard Widrow 开发的算法来设置权重用于加速训练。 该技术不是经常奏效，在某些场景下比纯粹的随机初始化来得更低效。
|
|
</p>
|
|
<p class="para">
|
|
该算法要求获取输入数据的范围(比如 最大和最小输入)，因此接受别的参数，数据(将会在网络中训练的数据)。
|
|
</p>
|
|
</div>
|
|
|
|
|
|
<div class="refsect1 parameters" id="refsect1-function.fann-init-weights-parameters">
|
|
<h3 class="title">参数</h3>
|
|
<dl>
|
|
|
|
|
|
<dt>
|
|
<code class="parameter">ann</code></dt>
|
|
|
|
<dd>
|
|
|
|
<p class="para">Neural network <span class="type"><a href="language.types.resource.html" class="type resource">resource</a></span>.</p>
|
|
</dd>
|
|
|
|
|
|
|
|
<dt>
|
|
<code class="parameter">train_data</code></dt>
|
|
|
|
<dd>
|
|
|
|
<p class="para">Neural network training data <span class="type"><a href="language.types.resource.html" class="type resource">resource</a></span>.</p>
|
|
</dd>
|
|
|
|
|
|
</dl>
|
|
|
|
</div>
|
|
|
|
|
|
<div class="refsect1 returnvalues" id="refsect1-function.fann-init-weights-returnvalues">
|
|
<h3 class="title">返回值</h3>
|
|
<p class="para">Returns <strong><code>TRUE</code></strong> on success, or <strong><code>FALSE</code></strong> otherwise.</p>
|
|
</div>
|
|
|
|
|
|
<div class="refsect1 seealso" id="refsect1-function.fann-init-weights-seealso">
|
|
<h3 class="title">参见</h3>
|
|
<p class="para">
|
|
<ul class="simplelist">
|
|
<li class="member"><span class="function"><a href="fann_randomize_weights.html" class="function" rel="rdfs-seeAlso">fann_randomize_weights()</a> - 给每个连接赋一个介于 min_weight 和 max_weight 之间的随机权重。</span></li>
|
|
<li class="member"><span class="function"><a href="fann_read_train_from_file.html" class="function" rel="rdfs-seeAlso">fann_read_train_from_file()</a> - 读取存储训练数据的文件。</span></li>
|
|
</ul>
|
|
</p>
|
|
</div>
|
|
|
|
|
|
|
|
</div></div></div></body></html> |