| <!DOCTYPE html> |
| |
| <html lang="en"> |
| <head> |
| <meta charset="utf-8"/> |
| <meta content="IE=edge" http-equiv="X-UA-Compatible"/> |
| <meta content="width=device-width, initial-scale=1" name="viewport"/> |
| <meta content="Gluon Neural Network Layers" property="og:title"> |
| <meta content="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/og-logo.png" property="og:image"> |
| <meta content="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/og-logo.png" property="og:image:secure_url"> |
| <meta content="Gluon Neural Network Layers" property="og:description"/> |
| <title>Gluon Neural Network Layers — mxnet documentation</title> |
| <link crossorigin="anonymous" href="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.6/css/bootstrap.min.css" integrity="sha384-1q8mTJOASx8j1Au+a5WDVnPi2lkFfwwEAa8hDDdjZlpLegxhjVME1fgjWPGmkzs7" rel="stylesheet"/> |
| <link href="https://maxcdn.bootstrapcdn.com/font-awesome/4.5.0/css/font-awesome.min.css" rel="stylesheet"/> |
| <link href="../../../_static/basic.css" rel="stylesheet" type="text/css"> |
| <link href="../../../_static/pygments.css" rel="stylesheet" type="text/css"> |
| <link href="../../../_static/mxnet.css" rel="stylesheet" type="text/css"/> |
| <script type="text/javascript"> |
| var DOCUMENTATION_OPTIONS = { |
| URL_ROOT: '../../../', |
| VERSION: '', |
| COLLAPSE_INDEX: false, |
| FILE_SUFFIX: '.html', |
| HAS_SOURCE: true, |
| SOURCELINK_SUFFIX: '.txt' |
| }; |
| </script> |
| <script src="https://code.jquery.com/jquery-1.11.1.min.js" type="text/javascript"></script> |
| <script src="../../../_static/underscore.js" type="text/javascript"></script> |
| <script src="../../../_static/searchtools_custom.js" type="text/javascript"></script> |
| <script src="../../../_static/doctools.js" type="text/javascript"></script> |
| <script src="../../../_static/selectlang.js" type="text/javascript"></script> |
| <script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.1/MathJax.js?config=TeX-AMS-MML_HTMLorMML" type="text/javascript"></script> |
| <script type="text/javascript"> jQuery(function() { Search.loadIndex("/versions/1.2.1/searchindex.js"); Search.init();}); </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> --> |
| <!-- --> |
| <!-- <script type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.0/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script> --> |
| <!-- --> |
| <link href="../../../genindex.html" rel="index" title="Index"> |
| <link href="../../../search.html" rel="search" title="Search"/> |
| <link href="gluon.html" rel="up" title="Gluon Package"/> |
| <link href="rnn.html" rel="next" title="Gluon Recurrent Neural Network API"/> |
| <link href="gluon.html" rel="prev" title="Gluon Package"/> |
| <link href="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/mxnet-icon.png" rel="icon" type="image/png"/> |
| </link></link></link></meta></meta></meta></head> |
| <body background="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/mxnet-background-compressed.jpeg" role="document"> |
| <div class="content-block"><div class="navbar navbar-fixed-top"> |
| <div class="container" id="navContainer"> |
| <div class="innder" id="header-inner"> |
| <h1 id="logo-wrap"> |
| <a href="../../../" id="logo"><img src="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/mxnet_logo.png"/></a> |
| </h1> |
| <nav class="nav-bar" id="main-nav"> |
| <a class="main-nav-link" href="/versions/1.2.1/install/index.html">Install</a> |
| <span id="dropdown-menu-position-anchor"> |
| <a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">Gluon <span class="caret"></span></a> |
| <ul class="dropdown-menu navbar-menu" id="package-dropdown-menu"> |
| <li><a class="main-nav-link" href="/versions/1.2.1/tutorials/gluon/gluon.html">About</a></li> |
| <li><a class="main-nav-link" href="https://www.d2l.ai/">Dive into Deep Learning</a></li> |
| <li><a class="main-nav-link" href="https://gluon-cv.mxnet.io">GluonCV Toolkit</a></li> |
| <li><a class="main-nav-link" href="https://gluon-nlp.mxnet.io/">GluonNLP Toolkit</a></li> |
| </ul> |
| </span> |
| <span id="dropdown-menu-position-anchor"> |
| <a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">API <span class="caret"></span></a> |
| <ul class="dropdown-menu navbar-menu" id="package-dropdown-menu"> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/python/index.html">Python</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/c++/index.html">C++</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/julia/index.html">Julia</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/perl/index.html">Perl</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/r/index.html">R</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/scala/index.html">Scala</a></li> |
| </ul> |
| </span> |
| <span id="dropdown-menu-position-anchor-docs"> |
| <a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">Docs <span class="caret"></span></a> |
| <ul class="dropdown-menu navbar-menu" id="package-dropdown-menu-docs"> |
| <li><a class="main-nav-link" href="/versions/1.2.1/faq/index.html">FAQ</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/tutorials/index.html">Tutorials</a> |
| <li><a class="main-nav-link" href="https://github.com/apache/incubator-mxnet/tree/1.2.1/example">Examples</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/architecture/index.html">Architecture</a></li> |
| <li><a class="main-nav-link" href="https://cwiki.apache.org/confluence/display/MXNET/Apache+MXNet+Home">Developer Wiki</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/model_zoo/index.html">Model Zoo</a></li> |
| <li><a class="main-nav-link" href="https://github.com/onnx/onnx-mxnet">ONNX</a></li> |
| </li></ul> |
| </span> |
| <span id="dropdown-menu-position-anchor-community"> |
| <a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">Community <span class="caret"></span></a> |
| <ul class="dropdown-menu navbar-menu" id="package-dropdown-menu-community"> |
| <li><a class="main-nav-link" href="http://discuss.mxnet.io">Forum</a></li> |
| <li><a class="main-nav-link" href="https://github.com/apache/incubator-mxnet/tree/1.2.1">Github</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/community/contribute.html">Contribute</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/community/powered_by.html">Powered By</a></li> |
| </ul> |
| </span> |
| <span id="dropdown-menu-position-anchor-version" style="position: relative"><a href="#" class="main-nav-link dropdown-toggle" data-toggle="dropdown" role="button" aria-haspopup="true" aria-expanded="true">1.2.1<span class="caret"></span></a><ul id="package-dropdown-menu" class="dropdown-menu"><li><a href="/">master</a></li><li><a href="/versions/1.7.0/">1.7.0</a></li><li><a href=/versions/1.6.0/>1.6.0</a></li><li><a href=/versions/1.5.0/>1.5.0</a></li><li><a href=/versions/1.4.1/>1.4.1</a></li><li><a href=/versions/1.3.1/>1.3.1</a></li><li><a href=/versions/1.2.1/>1.2.1</a></li><li><a href=/versions/1.1.0/>1.1.0</a></li><li><a href=/versions/1.0.0/>1.0.0</a></li><li><a href=/versions/0.12.1/>0.12.1</a></li><li><a href=/versions/0.11.0/>0.11.0</a></li></ul></span></nav> |
| <script> function getRootPath(){ return "../../../" } </script> |
| <div class="burgerIcon dropdown"> |
| <a class="dropdown-toggle" data-toggle="dropdown" href="#" role="button">☰</a> |
| <ul class="dropdown-menu" id="burgerMenu"> |
| <li><a href="/versions/1.2.1/install/index.html">Install</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/tutorials/index.html">Tutorials</a></li> |
| <li class="dropdown-submenu dropdown"> |
| <a aria-expanded="true" aria-haspopup="true" class="dropdown-toggle burger-link" data-toggle="dropdown" href="#" tabindex="-1">Gluon</a> |
| <ul class="dropdown-menu navbar-menu" id="package-dropdown-menu"> |
| <li><a class="main-nav-link" href="/versions/1.2.1/tutorials/gluon/gluon.html">About</a></li> |
| <li><a class="main-nav-link" href="http://gluon.mxnet.io">The Straight Dope (Tutorials)</a></li> |
| <li><a class="main-nav-link" href="https://gluon-cv.mxnet.io">GluonCV Toolkit</a></li> |
| <li><a class="main-nav-link" href="https://gluon-nlp.mxnet.io/">GluonNLP Toolkit</a></li> |
| </ul> |
| </li> |
| <li class="dropdown-submenu"> |
| <a aria-expanded="true" aria-haspopup="true" class="dropdown-toggle burger-link" data-toggle="dropdown" href="#" tabindex="-1">API</a> |
| <ul class="dropdown-menu"> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/python/index.html">Python</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/c++/index.html">C++</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/julia/index.html">Julia</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/perl/index.html">Perl</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/r/index.html">R</a></li> |
| <li><a class="main-nav-link" href="/versions/1.2.1/api/scala/index.html">Scala</a></li> |
| </ul> |
| </li> |
| <li class="dropdown-submenu"> |
| <a aria-expanded="true" aria-haspopup="true" class="dropdown-toggle burger-link" data-toggle="dropdown" href="#" tabindex="-1">Docs</a> |
| <ul class="dropdown-menu"> |
| <li><a href="/versions/1.2.1/faq/index.html" tabindex="-1">FAQ</a></li> |
| <li><a href="/versions/1.2.1/tutorials/index.html" tabindex="-1">Tutorials</a></li> |
| <li><a href="https://github.com/apache/incubator-mxnet/tree/1.2.1/example" tabindex="-1">Examples</a></li> |
| <li><a href="/versions/1.2.1/architecture/index.html" tabindex="-1">Architecture</a></li> |
| <li><a href="https://cwiki.apache.org/confluence/display/MXNET/Apache+MXNet+Home" tabindex="-1">Developer Wiki</a></li> |
| <li><a href="/versions/1.2.1/model_zoo/index.html" tabindex="-1">Gluon Model Zoo</a></li> |
| <li><a href="https://github.com/onnx/onnx-mxnet" tabindex="-1">ONNX</a></li> |
| </ul> |
| </li> |
| <li class="dropdown-submenu dropdown"> |
| <a aria-haspopup="true" class="dropdown-toggle burger-link" data-toggle="dropdown" href="#" role="button" tabindex="-1">Community</a> |
| <ul class="dropdown-menu"> |
| <li><a href="http://discuss.mxnet.io" tabindex="-1">Forum</a></li> |
| <li><a href="https://github.com/apache/incubator-mxnet/tree/1.2.1" tabindex="-1">Github</a></li> |
| <li><a href="/versions/1.2.1/community/contribute.html" tabindex="-1">Contribute</a></li> |
| <li><a href="/versions/1.2.1/community/powered_by.html" tabindex="-1">Powered By</a></li> |
| </ul> |
| </li> |
| <li id="dropdown-menu-position-anchor-version-mobile" class="dropdown-submenu" style="position: relative"><a href="#" tabindex="-1">1.2.1</a><ul class="dropdown-menu"><li><a tabindex="-1" href=/>master</a></li><li><a tabindex="-1" href=/versions/1.6.0/>1.6.0</a></li><li><a tabindex="-1" href=/versions/1.5.0/>1.5.0</a></li><li><a tabindex="-1" href=/versions/1.4.1/>1.4.1</a></li><li><a tabindex="-1" href=/versions/1.3.1/>1.3.1</a></li><li><a tabindex="-1" href=/versions/1.2.1/>1.2.1</a></li><li><a tabindex="-1" href=/versions/1.1.0/>1.1.0</a></li><li><a tabindex="-1" href=/versions/1.0.0/>1.0.0</a></li><li><a tabindex="-1" href=/versions/0.12.1/>0.12.1</a></li><li><a tabindex="-1" href=/versions/0.11.0/>0.11.0</a></li></ul></li></ul> |
| </div> |
| <div class="plusIcon dropdown"> |
| <a class="dropdown-toggle" data-toggle="dropdown" href="#" role="button"><span aria-hidden="true" class="glyphicon glyphicon-plus"></span></a> |
| <ul class="dropdown-menu dropdown-menu-right" id="plusMenu"></ul> |
| </div> |
| <div id="search-input-wrap"> |
| <form action="../../../search.html" autocomplete="off" class="" method="get" role="search"> |
| <div class="form-group inner-addon left-addon"> |
| <i class="glyphicon glyphicon-search"></i> |
| <input class="form-control" name="q" placeholder="Search" type="text"/> |
| </div> |
| <input name="check_keywords" type="hidden" value="yes"> |
| <input name="area" type="hidden" value="default"/> |
| </input></form> |
| <div id="search-preview"></div> |
| </div> |
| <div id="searchIcon"> |
| <span aria-hidden="true" class="glyphicon glyphicon-search"></span> |
| </div> |
| <!-- <div id="lang-select-wrap"> --> |
| <!-- <label id="lang-select-label"> --> |
| <!-- <\!-- <i class="fa fa-globe"></i> -\-> --> |
| <!-- <span></span> --> |
| <!-- </label> --> |
| <!-- <select id="lang-select"> --> |
| <!-- <option value="en">Eng</option> --> |
| <!-- <option value="zh">中文</option> --> |
| <!-- </select> --> |
| <!-- </div> --> |
| <!-- <a id="mobile-nav-toggle"> |
| <span class="mobile-nav-toggle-bar"></span> |
| <span class="mobile-nav-toggle-bar"></span> |
| <span class="mobile-nav-toggle-bar"></span> |
| </a> --> |
| </div> |
| </div> |
| </div> |
| <script type="text/javascript"> |
| $('body').css('background', 'white'); |
| </script> |
| <div class="container"> |
| <div class="row"> |
| <div aria-label="main navigation" class="sphinxsidebar leftsidebar" role="navigation"> |
| <div class="sphinxsidebarwrapper"> |
| <ul class="current"> |
| <li class="toctree-l1 current"><a class="reference internal" href="../index.html">Python Documents</a><ul class="current"> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#ndarray-api">NDArray API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#symbol-api">Symbol API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#module-api">Module API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#autograd-api">Autograd API</a></li> |
| <li class="toctree-l2 current"><a class="reference internal" href="../index.html#gluon-api">Gluon API</a><ul class="current"> |
| <li class="toctree-l3 current"><a class="reference internal" href="gluon.html">Gluon Package</a><ul class="current"> |
| <li class="toctree-l4 current"><a class="reference internal" href="gluon.html#overview">Overview</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="gluon.html#parameter">Parameter</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="gluon.html#containers">Containers</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="gluon.html#trainer">Trainer</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="gluon.html#utilities">Utilities</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="gluon.html#api-reference">API Reference</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3 current"><a class="current reference internal" href="#">Gluon Neural Network Layers</a><ul> |
| <li class="toctree-l4"><a class="reference internal" href="#overview">Overview</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="#basic-layers">Basic Layers</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="#convolutional-layers">Convolutional Layers</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="#pooling-layers">Pooling Layers</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="#activation-layers">Activation Layers</a></li> |
| <li class="toctree-l4"><a class="reference internal" href="#api-reference">API Reference</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="rnn.html">Gluon Recurrent Neural Network API</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="loss.html">Gluon Loss API</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="data.html">Gluon Data API</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="model_zoo.html">Gluon Model Zoo</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="contrib.html">Gluon Contrib API</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#kvstore-api">KVStore API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#io-api">IO API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#image-api">Image API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#optimization-api">Optimization API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#callback-api">Callback API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#metric-api">Metric API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#run-time-compilation-api">Run-Time Compilation API</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../index.html#contrib-package">Contrib Package</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l1"><a class="reference internal" href="../../r/index.html">R Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../julia/index.html">Julia Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../c++/index.html">C++ Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../scala/index.html">Scala Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../perl/index.html">Perl Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../faq/index.html">HowTo Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../architecture/index.html">System Documents</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../tutorials/index.html">Tutorials</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../community/index.html">Community</a></li> |
| </ul> |
| </div> |
| </div> |
| <div class="content"> |
| <div class="page-tracker"></div> |
| <div class="section" id="gluon-neural-network-layers"> |
| <span id="gluon-neural-network-layers"></span><h1>Gluon Neural Network Layers<a class="headerlink" href="#gluon-neural-network-layers" title="Permalink to this headline">¶</a></h1> |
| <div class="section" id="overview"> |
| <span id="overview"></span><h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2> |
| <p>This document lists the neural network blocks in Gluon:</p> |
| </div> |
| <div class="section" id="basic-layers"> |
| <span id="basic-layers"></span><h2>Basic Layers<a class="headerlink" href="#basic-layers" title="Permalink to this headline">¶</a></h2> |
| <table border="1" class="longtable docutils"> |
| <colgroup> |
| <col width="10%"/> |
| <col width="90%"/> |
| </colgroup> |
| <tbody valign="top"> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.Dense" title="mxnet.gluon.nn.Dense"><code class="xref py py-obj docutils literal"><span class="pre">Dense</span></code></a></td> |
| <td>Just your regular densely-connected NN layer.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.Dropout" title="mxnet.gluon.nn.Dropout"><code class="xref py py-obj docutils literal"><span class="pre">Dropout</span></code></a></td> |
| <td>Applies Dropout to the input.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.BatchNorm" title="mxnet.gluon.nn.BatchNorm"><code class="xref py py-obj docutils literal"><span class="pre">BatchNorm</span></code></a></td> |
| <td>Batch normalization layer (Ioffe and Szegedy, 2014).</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.InstanceNorm" title="mxnet.gluon.nn.InstanceNorm"><code class="xref py py-obj docutils literal"><span class="pre">InstanceNorm</span></code></a></td> |
| <td>Applies instance normalization to the n-dimensional input array.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.LayerNorm" title="mxnet.gluon.nn.LayerNorm"><code class="xref py py-obj docutils literal"><span class="pre">LayerNorm</span></code></a></td> |
| <td>Applies layer normalization to the n-dimensional input array.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.Embedding" title="mxnet.gluon.nn.Embedding"><code class="xref py py-obj docutils literal"><span class="pre">Embedding</span></code></a></td> |
| <td>Turns non-negative integers (indexes/tokens) into dense vectors of fixed size.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.Flatten" title="mxnet.gluon.nn.Flatten"><code class="xref py py-obj docutils literal"><span class="pre">Flatten</span></code></a></td> |
| <td>Flattens the input to two dimensional.</td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="convolutional-layers"> |
| <span id="convolutional-layers"></span><h2>Convolutional Layers<a class="headerlink" href="#convolutional-layers" title="Permalink to this headline">¶</a></h2> |
| <table border="1" class="longtable docutils"> |
| <colgroup> |
| <col width="10%"/> |
| <col width="90%"/> |
| </colgroup> |
| <tbody valign="top"> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.Conv1D" title="mxnet.gluon.nn.Conv1D"><code class="xref py py-obj docutils literal"><span class="pre">Conv1D</span></code></a></td> |
| <td>1D convolution layer (e.g. temporal convolution).</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.Conv2D" title="mxnet.gluon.nn.Conv2D"><code class="xref py py-obj docutils literal"><span class="pre">Conv2D</span></code></a></td> |
| <td>2D convolution layer (e.g. spatial convolution over images).</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.Conv3D" title="mxnet.gluon.nn.Conv3D"><code class="xref py py-obj docutils literal"><span class="pre">Conv3D</span></code></a></td> |
| <td>3D convolution layer (e.g. spatial convolution over volumes).</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.Conv1DTranspose" title="mxnet.gluon.nn.Conv1DTranspose"><code class="xref py py-obj docutils literal"><span class="pre">Conv1DTranspose</span></code></a></td> |
| <td>Transposed 1D convolution layer (sometimes called Deconvolution).</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.Conv2DTranspose" title="mxnet.gluon.nn.Conv2DTranspose"><code class="xref py py-obj docutils literal"><span class="pre">Conv2DTranspose</span></code></a></td> |
| <td>Transposed 2D convolution layer (sometimes called Deconvolution).</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.Conv3DTranspose" title="mxnet.gluon.nn.Conv3DTranspose"><code class="xref py py-obj docutils literal"><span class="pre">Conv3DTranspose</span></code></a></td> |
| <td>Transposed 3D convolution layer (sometimes called Deconvolution).</td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="pooling-layers"> |
| <span id="pooling-layers"></span><h2>Pooling Layers<a class="headerlink" href="#pooling-layers" title="Permalink to this headline">¶</a></h2> |
| <table border="1" class="longtable docutils"> |
| <colgroup> |
| <col width="10%"/> |
| <col width="90%"/> |
| </colgroup> |
| <tbody valign="top"> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.MaxPool1D" title="mxnet.gluon.nn.MaxPool1D"><code class="xref py py-obj docutils literal"><span class="pre">MaxPool1D</span></code></a></td> |
| <td>Max pooling operation for one dimensional data.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.MaxPool2D" title="mxnet.gluon.nn.MaxPool2D"><code class="xref py py-obj docutils literal"><span class="pre">MaxPool2D</span></code></a></td> |
| <td>Max pooling operation for two dimensional (spatial) data.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.MaxPool3D" title="mxnet.gluon.nn.MaxPool3D"><code class="xref py py-obj docutils literal"><span class="pre">MaxPool3D</span></code></a></td> |
| <td>Max pooling operation for 3D data (spatial or spatio-temporal).</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.AvgPool1D" title="mxnet.gluon.nn.AvgPool1D"><code class="xref py py-obj docutils literal"><span class="pre">AvgPool1D</span></code></a></td> |
| <td>Average pooling operation for temporal data.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.AvgPool2D" title="mxnet.gluon.nn.AvgPool2D"><code class="xref py py-obj docutils literal"><span class="pre">AvgPool2D</span></code></a></td> |
| <td>Average pooling operation for spatial data.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.AvgPool3D" title="mxnet.gluon.nn.AvgPool3D"><code class="xref py py-obj docutils literal"><span class="pre">AvgPool3D</span></code></a></td> |
| <td>Average pooling operation for 3D data (spatial or spatio-temporal).</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool1D" title="mxnet.gluon.nn.GlobalMaxPool1D"><code class="xref py py-obj docutils literal"><span class="pre">GlobalMaxPool1D</span></code></a></td> |
| <td>Global max pooling operation for temporal data.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool2D" title="mxnet.gluon.nn.GlobalMaxPool2D"><code class="xref py py-obj docutils literal"><span class="pre">GlobalMaxPool2D</span></code></a></td> |
| <td>Global max pooling operation for spatial data.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.GlobalMaxPool3D" title="mxnet.gluon.nn.GlobalMaxPool3D"><code class="xref py py-obj docutils literal"><span class="pre">GlobalMaxPool3D</span></code></a></td> |
| <td>Global max pooling operation for 3D data.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool1D" title="mxnet.gluon.nn.GlobalAvgPool1D"><code class="xref py py-obj docutils literal"><span class="pre">GlobalAvgPool1D</span></code></a></td> |
| <td>Global average pooling operation for temporal data.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool2D" title="mxnet.gluon.nn.GlobalAvgPool2D"><code class="xref py py-obj docutils literal"><span class="pre">GlobalAvgPool2D</span></code></a></td> |
| <td>Global average pooling operation for spatial data.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.GlobalAvgPool3D" title="mxnet.gluon.nn.GlobalAvgPool3D"><code class="xref py py-obj docutils literal"><span class="pre">GlobalAvgPool3D</span></code></a></td> |
| <td>Global max pooling operation for 3D data.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.ReflectionPad2D" title="mxnet.gluon.nn.ReflectionPad2D"><code class="xref py py-obj docutils literal"><span class="pre">ReflectionPad2D</span></code></a></td> |
| <td>Pads the input tensor using the reflection of the input boundary.</td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="activation-layers"> |
| <span id="activation-layers"></span><h2>Activation Layers<a class="headerlink" href="#activation-layers" title="Permalink to this headline">¶</a></h2> |
| <table border="1" class="longtable docutils"> |
| <colgroup> |
| <col width="10%"/> |
| <col width="90%"/> |
| </colgroup> |
| <tbody valign="top"> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.Activation" title="mxnet.gluon.nn.Activation"><code class="xref py py-obj docutils literal"><span class="pre">Activation</span></code></a></td> |
| <td>Applies an activation function to input.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.LeakyReLU" title="mxnet.gluon.nn.LeakyReLU"><code class="xref py py-obj docutils literal"><span class="pre">LeakyReLU</span></code></a></td> |
| <td>Leaky version of a Rectified Linear Unit.</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.PReLU" title="mxnet.gluon.nn.PReLU"><code class="xref py py-obj docutils literal"><span class="pre">PReLU</span></code></a></td> |
| <td>Parametric leaky version of a Rectified Linear Unit.</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.ELU" title="mxnet.gluon.nn.ELU"><code class="xref py py-obj docutils literal"><span class="pre">ELU</span></code></a></td> |
| <td>Exponential Linear Unit (ELU)</td> |
| </tr> |
| <tr class="row-odd"><td><a class="reference internal" href="#mxnet.gluon.nn.SELU" title="mxnet.gluon.nn.SELU"><code class="xref py py-obj docutils literal"><span class="pre">SELU</span></code></a></td> |
| <td>Scaled Exponential Linear Unit (SELU)</td> |
| </tr> |
| <tr class="row-even"><td><a class="reference internal" href="#mxnet.gluon.nn.Swish" title="mxnet.gluon.nn.Swish"><code class="xref py py-obj docutils literal"><span class="pre">Swish</span></code></a></td> |
| <td>Swish Activation function</td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| <div class="section" id="api-reference"> |
| <span id="api-reference"></span><h2>API Reference<a class="headerlink" href="#api-reference" title="Permalink to this headline">¶</a></h2> |
| <script src="../../_static/js/auto_module_index.js" type="text/javascript"></script><span class="target" id="module-mxnet.gluon.nn"></span><p>Neural network layers.</p> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Activation"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Activation</code><span class="sig-paren">(</span><em>activation</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#Activation"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Activation" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies an activation function to input.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><strong>activation</strong> (<em>str</em>) – Name of activation function to use. |
| See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a> for available choices.</td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.AvgPool1D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">AvgPool1D</code><span class="sig-paren">(</span><em>pool_size=2</em>, <em>strides=None</em>, <em>padding=0</em>, <em>layout='NCW'</em>, <em>ceil_mode=False</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#AvgPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.AvgPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Average pooling operation for temporal data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>pool_size</strong> (<em>int</em>) – Size of the max pooling windows.</li> |
| <li><strong>strides</strong> (<em>int</em><em>, or </em><em>None</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</li> |
| <li><strong>padding</strong> (<em>int</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and weight. Can be ‘NCW’, ‘NWC’, etc. |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. padding is applied on ‘W’ dimension.</li> |
| <li><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">-</span><span class="n">pool_size</span><span class="p">)</span><span class="o">/</span><span class="n">strides</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.AvgPool2D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">AvgPool2D</code><span class="sig-paren">(</span><em>pool_size=(2</em>, <em>2)</em>, <em>strides=None</em>, <em>padding=0</em>, <em>ceil_mode=False</em>, <em>layout='NCHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#AvgPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.AvgPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Average pooling operation for spatial data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – Size of the max pooling windows.</li> |
| <li><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 2 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and weight. Can be ‘NCHW’, ‘NHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. padding is applied on ‘H’ and ‘W’ dimension.</li> |
| <li><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When True, will use ceil instead of floor to compute the output shape.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.AvgPool3D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">AvgPool3D</code><span class="sig-paren">(</span><em>pool_size=(2</em>, <em>2</em>, <em>2)</em>, <em>strides=None</em>, <em>padding=0</em>, <em>ceil_mode=False</em>, <em>layout='NCDHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#AvgPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.AvgPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Average pooling operation for 3D data (spatial or spatio-temporal).</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – Size of the max pooling windows.</li> |
| <li><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 3 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and weight. Can be ‘NCDHW’, ‘NDHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. padding is applied on ‘D’, ‘H’ and ‘W’ |
| dimension.</li> |
| <li><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When True, will use ceil instead of floor to compute the output shape.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">depth</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True,</cite> ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.BatchNorm"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">BatchNorm</code><span class="sig-paren">(</span><em>axis=1</em>, <em>momentum=0.9</em>, <em>epsilon=1e-05</em>, <em>center=True</em>, <em>scale=True</em>, <em>use_global_stats=False</em>, <em>beta_initializer='zeros'</em>, <em>gamma_initializer='ones'</em>, <em>running_mean_initializer='zeros'</em>, <em>running_variance_initializer='ones'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#BatchNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.BatchNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Batch normalization layer (Ioffe and Szegedy, 2014). |
| Normalizes the input at each batch, i.e. applies a transformation |
| that maintains the mean activation close to 0 and the activation |
| standard deviation close to 1.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>axis</strong> (<em>int</em><em>, </em><em>default 1</em>) – The axis that should be normalized. This is typically the channels |
| (C) axis. For instance, after a <cite>Conv2D</cite> layer with <cite>layout=’NCHW’</cite>, |
| set <cite>axis=1</cite> in <cite>BatchNorm</cite>. If <cite>layout=’NHWC’</cite>, then set <cite>axis=3</cite>.</li> |
| <li><strong>momentum</strong> (<em>float</em><em>, </em><em>default 0.9</em>) – Momentum for the moving average.</li> |
| <li><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</li> |
| <li><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</li> |
| <li><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used. |
| When the next layer is linear (also e.g. <cite>nn.relu</cite>), |
| this can be disabled since the scaling |
| will be done by the next layer.</li> |
| <li><strong>use_global_stats</strong> (<em>bool</em><em>, </em><em>default False</em>) – If True, use global moving statistics instead of local batch-norm. This will force |
| change batch-norm into a scale shift operator. |
| If False, use local batch-norm.</li> |
| <li><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</li> |
| <li><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</li> |
| <li><strong>moving_mean_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the moving mean.</li> |
| <li><strong>moving_variance_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the moving variance.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv1D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Conv1D</code><span class="sig-paren">(</span><em>channels</em>, <em>kernel_size</em>, <em>strides=1</em>, <em>padding=0</em>, <em>dilation=1</em>, <em>groups=1</em>, <em>layout='NCW'</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>1D convolution layer (e.g. temporal convolution).</p> |
| <p>This layer creates a convolution kernel that is convolved |
| with the layer input over a single spatial (or temporal) dimension |
| to produce a tensor of outputs. |
| If <cite>use_bias</cite> is True, a bias vector is created and added to the outputs. |
| Finally, if <cite>activation</cite> is not <cite>None</cite>, |
| it is applied to the outputs as well.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</li> |
| <li><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Specifies the dimensions of the convolution window.</li> |
| <li><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em><em>,</em>) – Specify the strides of the convolution.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 1 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</li> |
| <li><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 1 int</em>) – Specifies the dilation rate to use for dilated convolution.</li> |
| <li><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and weight. Can be ‘NCW’, ‘NWC’, etc. |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Convolution is applied on the ‘W’ dimension.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">-</span><span class="n">dilation</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv1DTranspose"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Conv1DTranspose</code><span class="sig-paren">(</span><em>channels</em>, <em>kernel_size</em>, <em>strides=1</em>, <em>padding=0</em>, <em>output_padding=0</em>, <em>dilation=1</em>, <em>groups=1</em>, <em>layout='NCW'</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv1DTranspose"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv1DTranspose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Transposed 1D convolution layer (sometimes called Deconvolution).</p> |
| <p>The need for transposed convolutions generally arises |
| from the desire to use a transformation going in the opposite direction |
| of a normal convolution, i.e., from something that has the shape of the |
| output of some convolution to something that has the shape of its input |
| while maintaining a connectivity pattern that is compatible with |
| said convolution.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</li> |
| <li><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dimensions of the convolution window.</li> |
| <li><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em><em>,</em>) – Specify the strides of the convolution.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</li> |
| <li><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dilation rate to use for dilated convolution.</li> |
| <li><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and weight. Can be ‘NCW’, ‘NWC’, etc. |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Convolution is applied on the ‘W’ dimension.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="p">(</span><span class="n">width</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">+</span><span class="n">kernel_size</span><span class="o">+</span><span class="n">output_padding</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv2D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Conv2D</code><span class="sig-paren">(</span><em>channels</em>, <em>kernel_size</em>, <em>strides=(1</em>, <em>1)</em>, <em>padding=(0</em>, <em>0)</em>, <em>dilation=(1</em>, <em>1)</em>, <em>groups=1</em>, <em>layout='NCHW'</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>2D convolution layer (e.g. spatial convolution over images).</p> |
| <p>This layer creates a convolution kernel that is convolved |
| with the layer input to produce a tensor of |
| outputs. If <cite>use_bias</cite> is True, |
| a bias vector is created and added to the outputs. Finally, if |
| <cite>activation</cite> is not <cite>None</cite>, it is applied to the outputs as well.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</li> |
| <li><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Specifies the dimensions of the convolution window.</li> |
| <li><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em><em>,</em>) – Specify the strides of the convolution.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 2 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</li> |
| <li><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 2 int</em>) – Specifies the dilation rate to use for dilated convolution.</li> |
| <li><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and weight. Can be ‘NCHW’, ‘NHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. Convolution is applied on the ‘H’ and |
| ‘W’ dimensions.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</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">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</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">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv2DTranspose"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Conv2DTranspose</code><span class="sig-paren">(</span><em>channels</em>, <em>kernel_size</em>, <em>strides=(1</em>, <em>1)</em>, <em>padding=(0</em>, <em>0)</em>, <em>output_padding=(0</em>, <em>0)</em>, <em>dilation=(1</em>, <em>1)</em>, <em>groups=1</em>, <em>layout='NCHW'</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv2DTranspose"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv2DTranspose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Transposed 2D convolution layer (sometimes called Deconvolution).</p> |
| <p>The need for transposed convolutions generally arises |
| from the desire to use a transformation going in the opposite direction |
| of a normal convolution, i.e., from something that has the shape of the |
| output of some convolution to something that has the shape of its input |
| while maintaining a connectivity pattern that is compatible with |
| said convolution.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</li> |
| <li><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dimensions of the convolution window.</li> |
| <li><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em><em>,</em>) – Specify the strides of the convolution.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</li> |
| <li><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dilation rate to use for dilated convolution.</li> |
| <li><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and weight. Can be ‘NCHW’, ‘NHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. Convolution is applied on the ‘H’ and |
| ‘W’ dimensions.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="p">(</span><span class="n">height</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="p">(</span><span class="n">width</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv3D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Conv3D</code><span class="sig-paren">(</span><em>channels</em>, <em>kernel_size</em>, <em>strides=(1</em>, <em>1</em>, <em>1)</em>, <em>padding=(0</em>, <em>0</em>, <em>0)</em>, <em>dilation=(1</em>, <em>1</em>, <em>1)</em>, <em>groups=1</em>, <em>layout='NCDHW'</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>3D convolution layer (e.g. spatial convolution over volumes).</p> |
| <p>This layer creates a convolution kernel that is convolved |
| with the layer input to produce a tensor of |
| outputs. If <cite>use_bias</cite> is <cite>True</cite>, |
| a bias vector is created and added to the outputs. Finally, if |
| <cite>activation</cite> is not <cite>None</cite>, it is applied to the outputs as well.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</li> |
| <li><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dimensions of the convolution window.</li> |
| <li><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em><em>,</em>) – Specify the strides of the convolution.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</li> |
| <li><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dilation rate to use for dilated convolution.</li> |
| <li><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and weight. Can be ‘NCDHW’, ‘NDHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. Convolution is applied on the ‘D’, |
| ‘H’ and ‘W’ dimensions.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">depth</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</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">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</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">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="n">dilation</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">*</span><span class="p">(</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">2</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">1</span><span class="p">)</span><span class="o">/</span><span class="n">stride</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Conv3DTranspose"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Conv3DTranspose</code><span class="sig-paren">(</span><em>channels</em>, <em>kernel_size</em>, <em>strides=(1</em>, <em>1</em>, <em>1)</em>, <em>padding=(0</em>, <em>0</em>, <em>0)</em>, <em>output_padding=(0</em>, <em>0</em>, <em>0)</em>, <em>dilation=(1</em>, <em>1</em>, <em>1)</em>, <em>groups=1</em>, <em>layout='NCDHW'</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#Conv3DTranspose"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Conv3DTranspose" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Transposed 3D convolution layer (sometimes called Deconvolution).</p> |
| <p>The need for transposed convolutions generally arises |
| from the desire to use a transformation going in the opposite direction |
| of a normal convolution, i.e., from something that has the shape of the |
| output of some convolution to something that has the shape of its input |
| while maintaining a connectivity pattern that is compatible with |
| said convolution.</p> |
| <p>If <cite>in_channels</cite> is not specified, <cite>Parameter</cite> initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_channels</cite> will be |
| inferred from the shape of input data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>channels</strong> (<em>int</em>) – The dimensionality of the output space, i.e. the number of output |
| channels (filters) in the convolution.</li> |
| <li><strong>kernel_size</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dimensions of the convolution window.</li> |
| <li><strong>strides</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em><em>,</em>) – Specify the strides of the convolution.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>a tuple/list of 3 int</em><em>,</em>) – If padding is non-zero, then the input is implicitly zero-padded |
| on both sides for padding number of points</li> |
| <li><strong>dilation</strong> (<em>int</em><em> or </em><em>tuple/list of 3 int</em>) – Specifies the dilation rate to use for dilated convolution.</li> |
| <li><strong>groups</strong> (<em>int</em>) – Controls the connections between inputs and outputs. |
| At groups=1, all inputs are convolved to all outputs. |
| At groups=2, the operation becomes equivalent to having two conv |
| layers side by side, each seeing half the input channels, and producing |
| half the output channels, and both subsequently concatenated.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and weight. Can be ‘NCDHW’, ‘NDHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. Convolution is applied on the ‘D’, |
| ‘H’, and ‘W’ dimensions.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – The number of input channels to this layer. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See <a class="reference internal" href="../ndarray/ndarray.html#mxnet.ndarray.Activation" title="mxnet.ndarray.Activation"><code class="xref py py-func docutils literal"><span class="pre">Activation()</span></code></a>. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>weight</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="p">(</span><span class="n">depth</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="p">(</span><span class="n">height</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="p">(</span><span class="n">width</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">*</span><span class="n">strides</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">+</span><span class="n">kernel_size</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">+</span><span class="n">output_padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Dense"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Dense</code><span class="sig-paren">(</span><em>units</em>, <em>activation=None</em>, <em>use_bias=True</em>, <em>flatten=True</em>, <em>dtype='float32'</em>, <em>weight_initializer=None</em>, <em>bias_initializer='zeros'</em>, <em>in_units=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dense"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dense" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Just your regular densely-connected NN layer.</p> |
| <p><cite>Dense</cite> implements the operation: |
| <cite>output = activation(dot(input, weight) + bias)</cite> |
| where <cite>activation</cite> is the element-wise activation function |
| passed as the <cite>activation</cite> argument, <cite>weight</cite> is a weights matrix |
| created by the layer, and <cite>bias</cite> is a bias vector created by the layer |
| (only applicable if <cite>use_bias</cite> is <cite>True</cite>).</p> |
| <p>Note: the input must be a tensor with rank 2. Use <cite>flatten</cite> to convert it |
| to rank 2 manually if necessary.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>units</strong> (<em>int</em>) – Dimensionality of the output space.</li> |
| <li><strong>activation</strong> (<em>str</em>) – Activation function to use. See help on <cite>Activation</cite> layer. |
| If you don’t specify anything, no activation is applied |
| (ie. “linear” activation: <cite>a(x) = x</cite>).</li> |
| <li><strong>use_bias</strong> (<em>bool</em>) – Whether the layer uses a bias vector.</li> |
| <li><strong>flatten</strong> (<em>bool</em>) – Whether the input tensor should be flattened. |
| If true, all but the first axis of input data are collapsed together. |
| If false, all but the last axis of input data are kept the same, and the transformation |
| applies on the last axis.</li> |
| <li><strong>dtype</strong> (<em>str</em><em> or </em><em>np.dtype</em><em>, </em><em>default 'float32'</em>) – Data type of output embeddings.</li> |
| <li><strong>weight_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the <cite>kernel</cite> weights matrix.</li> |
| <li><strong>bias_initializer</strong> (str or <cite>Initializer</cite>) – Initializer for the bias vector.</li> |
| <li><strong>in_units</strong> (<em>int</em><em>, </em><em>optional</em>) – Size of the input data. If not specified, initialization will be |
| deferred to the first time <cite>forward</cite> is called and <cite>in_units</cite> |
| will be inferred from the shape of input data.</li> |
| <li><strong>prefix</strong> (<em>str</em><em> or </em><em>None</em>) – See document of <cite>Block</cite>.</li> |
| <li><strong>params</strong> (<a class="reference internal" href="gluon.html#mxnet.gluon.ParameterDict" title="mxnet.gluon.ParameterDict"><em>ParameterDict</em></a><em> or </em><em>None</em>) – See document of <cite>Block</cite>.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: if <cite>flatten</cite> is True, <cite>data</cite> should be a tensor with shape |
| <cite>(batch_size, x1, x2, ..., xn)</cite>, where x1 * x2 * ... * xn is equal to |
| <cite>in_units</cite>. If <cite>flatten</cite> is False, <cite>data</cite> should have shape |
| <cite>(x1, x2, ..., xn, in_units)</cite>.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: if <cite>flatten</cite> is True, <cite>out</cite> will be a tensor with shape |
| <cite>(batch_size, units)</cite>. If <cite>flatten</cite> is False, <cite>out</cite> will have shape |
| <cite>(x1, x2, ..., xn, units)</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Dropout"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Dropout</code><span class="sig-paren">(</span><em>rate</em>, <em>axes=()</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Dropout"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Dropout" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies Dropout to the input.</p> |
| <p>Dropout consists in randomly setting a fraction <cite>rate</cite> of input units |
| to 0 at each update during training time, which helps prevent overfitting.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>rate</strong> (<em>float</em>) – Fraction of the input units to drop. Must be a number between 0 and 1.</li> |
| <li><strong>axes</strong> (<em>tuple of int</em><em>, </em><em>default</em><em> (</em><em>)</em>) – The axes on which dropout mask is shared. If empty, regular dropout is applied.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="http://www.cs.toronto.edu/~rsalakhu/papers/srivastava14a.pdf">Dropout: A Simple Way to Prevent Neural Networks from Overfitting</a></p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.ELU"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">ELU</code><span class="sig-paren">(</span><em>alpha=1.0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#ELU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ELU" title="Permalink to this definition">¶</a></dt> |
| <dd><dl class="docutils"> |
| <dt>Exponential Linear Unit (ELU)</dt> |
| <dd>“Fast and Accurate Deep Network Learning by Exponential Linear Units”, Clevert et al, 2016 |
| <a class="reference external" href="https://arxiv.org/abs/1511.07289">https://arxiv.org/abs/1511.07289</a> |
| Published as a conference paper at ICLR 2016</dd> |
| </dl> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><strong>alpha</strong> (<em>float</em>) – The alpha parameter as described by Clevert et al, 2016</td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Embedding"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Embedding</code><span class="sig-paren">(</span><em>input_dim</em>, <em>output_dim</em>, <em>dtype='float32'</em>, <em>weight_initializer=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Embedding"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Embedding" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Turns non-negative integers (indexes/tokens) into dense vectors |
| of fixed size. eg. [4, 20] -> [[0.25, 0.1], [0.6, -0.2]]</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>input_dim</strong> (<em>int</em>) – Size of the vocabulary, i.e. maximum integer index + 1.</li> |
| <li><strong>output_dim</strong> (<em>int</em>) – Dimension of the dense embedding.</li> |
| <li><strong>dtype</strong> (<em>str</em><em> or </em><em>np.dtype</em><em>, </em><em>default 'float32'</em>) – Data type of output embeddings.</li> |
| <li><strong>weight_initializer</strong> (<a class="reference internal" href="../optimization/optimization.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Initializer for the <cite>embeddings</cite> matrix.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: (N-1)-D tensor with shape: <cite>(x1, x2, ..., xN-1)</cite>.</li> |
| </ul> |
| </dd> |
| <dt>Output:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: N-D tensor with shape: <cite>(x1, x2, ..., xN-1, output_dim)</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Flatten"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Flatten</code><span class="sig-paren">(</span><em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Flatten"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Flatten" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Flattens the input to two dimensional.</p> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape <cite>(N, x1, x2, ..., xn)</cite></li> |
| </ul> |
| </dd> |
| <dt>Output:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: 2D tensor with shape: <cite>(N, x1 cdot x2 cdot ... cdot xn)</cite></li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalAvgPool1D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">GlobalAvgPool1D</code><span class="sig-paren">(</span><em>layout='NCW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalAvgPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalAvgPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Global average pooling operation for temporal data.</p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalAvgPool2D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">GlobalAvgPool2D</code><span class="sig-paren">(</span><em>layout='NCHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalAvgPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalAvgPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Global average pooling operation for spatial data.</p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalAvgPool3D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">GlobalAvgPool3D</code><span class="sig-paren">(</span><em>layout='NCDHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalAvgPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalAvgPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Global max pooling operation for 3D data.</p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalMaxPool1D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">GlobalMaxPool1D</code><span class="sig-paren">(</span><em>layout='NCW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalMaxPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalMaxPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Global max pooling operation for temporal data.</p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalMaxPool2D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">GlobalMaxPool2D</code><span class="sig-paren">(</span><em>layout='NCHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalMaxPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalMaxPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Global max pooling operation for spatial data.</p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.GlobalMaxPool3D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">GlobalMaxPool3D</code><span class="sig-paren">(</span><em>layout='NCDHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#GlobalMaxPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.GlobalMaxPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Global max pooling operation for 3D data.</p> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.HybridLambda"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">HybridLambda</code><span class="sig-paren">(</span><em>function</em>, <em>prefix=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#HybridLambda"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.HybridLambda" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Wraps an operator or an expression as a HybridBlock object.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>function</strong> (<em>str</em><em> or </em><em>function</em>) – <p>Function used in lambda must be one of the following: |
| 1) the name of an operator that is available in both symbol and ndarray. For example:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">HybridLambda</span><span class="p">(</span><span class="s1">'tanh'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <ol class="arabic" start="2"> |
| <li>a function that conforms to “def function(F, data, <a href="#id1"><span class="problematic" id="id2">*</span></a>args)”. For example:<div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">HybridLambda</span><span class="p">(</span><span class="k">lambda</span> <span class="n">F</span><span class="p">,</span> <span class="n">x</span><span class="p">:</span> <span class="n">F</span><span class="o">.</span><span class="n">LeakyReLU</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">slope</span><span class="o">=</span><span class="mf">0.1</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| </li> |
| </ol> |
| </li> |
| <li><strong>Inputs</strong> – <ul> |
| <li>** <em>args *</em>: one or more input data. First argument must be symbol or ndarray.</li> |
| </ul> |
| <p>Their shapes depend on the function.</p> |
| </li> |
| <li><strong>Output</strong> – <ul> |
| <li>** <em>outputs *</em>: one or more output data. Their shapes depend on the function.</li> |
| </ul> |
| </li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.InstanceNorm"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">InstanceNorm</code><span class="sig-paren">(</span><em>axis=1</em>, <em>epsilon=1e-05</em>, <em>center=True</em>, <em>scale=False</em>, <em>beta_initializer='zeros'</em>, <em>gamma_initializer='ones'</em>, <em>in_channels=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#InstanceNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.InstanceNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies instance normalization to the n-dimensional input array. |
| This operator takes an n-dimensional input array where (n>2) and normalizes |
| the input using the following formula:</p> |
| <div class="math"> |
| \[ \begin{align}\begin{aligned}\bar{C} = \{i \mid i \neq 0, i \neq axis\}\\out = \frac{x - mean[data, \bar{C}]}{ \sqrt{Var[data, \bar{C}]} + \epsilon} |
| * gamma + beta\end{aligned}\end{align} \]</div> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>axis</strong> (<em>int</em><em>, </em><em>default 1</em>) – The axis that will be excluded in the normalization process. This is typically the channels |
| (C) axis. For instance, after a <cite>Conv2D</cite> layer with <cite>layout=’NCHW’</cite>, |
| set <cite>axis=1</cite> in <cite>InstanceNorm</cite>. If <cite>layout=’NHWC’</cite>, then set <cite>axis=3</cite>. Data will be |
| normalized along axes excluding the first axis and the axis given.</li> |
| <li><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</li> |
| <li><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</li> |
| <li><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used. |
| When the next layer is linear (also e.g. <cite>nn.relu</cite>), |
| this can be disabled since the scaling |
| will be done by the next layer.</li> |
| <li><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</li> |
| <li><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://arxiv.org/abs/1607.08022">Instance Normalization: The Missing Ingredient for Fast Stylization</a></p> |
| <p class="rubric">Examples</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="c1"># Input of shape (2,1,2)</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">array</span><span class="p">([[[</span> <span class="mf">1.1</span><span class="p">,</span> <span class="mf">2.2</span><span class="p">]],</span> |
| <span class="gp">... </span> <span class="p">[[</span> <span class="mf">3.3</span><span class="p">,</span> <span class="mf">4.4</span><span class="p">]]])</span> |
| <span class="gp">>>> </span><span class="c1"># Instance normalization is calculated with the above formula</span> |
| <span class="gp">>>> </span><span class="n">layer</span> <span class="o">=</span> <span class="n">InstanceNorm</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">ctx</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="go">[[[-0.99998355 0.99998331]]</span> |
| <span class="go"> [[-0.99998319 0.99998361]]]</span> |
| <span class="go"><NDArray 2x1x2 @cpu(0)></span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Lambda"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Lambda</code><span class="sig-paren">(</span><em>function</em>, <em>prefix=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#Lambda"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Lambda" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Wraps an operator or an expression as a Block object.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>function</strong> (<em>str</em><em> or </em><em>function</em>) – <p>Function used in lambda must be one of the following: |
| 1) the name of an operator that is available in ndarray. For example:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">Lambda</span><span class="p">(</span><span class="s1">'tanh'</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <ol class="arabic" start="2"> |
| <li>a function that conforms to “def function(<a href="#id3"><span class="problematic" id="id4">*</span></a>args)”. For example:<div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">block</span> <span class="o">=</span> <span class="n">Lambda</span><span class="p">(</span><span class="k">lambda</span> <span class="n">x</span><span class="p">:</span> <span class="n">nd</span><span class="o">.</span><span class="n">LeakyReLU</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">slope</span><span class="o">=</span><span class="mf">0.1</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| </li> |
| </ol> |
| </li> |
| <li><strong>Inputs</strong> – <ul> |
| <li>** <em>args *</em>: one or more input data. Their shapes depend on the function.</li> |
| </ul> |
| </li> |
| <li><strong>Output</strong> – <ul> |
| <li>** <em>outputs *</em>: one or more output data. Their shapes depend on the function.</li> |
| </ul> |
| </li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.LayerNorm"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">LayerNorm</code><span class="sig-paren">(</span><em>axis=-1</em>, <em>epsilon=1e-05</em>, <em>center=True</em>, <em>scale=True</em>, <em>beta_initializer='zeros'</em>, <em>gamma_initializer='ones'</em>, <em>in_channels=0</em>, <em>prefix=None</em>, <em>params=None</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/basic_layers.html#LayerNorm"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LayerNorm" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Applies layer normalization to the n-dimensional input array. |
| This operator takes an n-dimensional input array and normalizes |
| the input using the given axis:</p> |
| <div class="math"> |
| \[out = \frac{x - mean[data, axis]}{ \sqrt{Var[data, axis]} + \epsilon} * gamma + beta\]</div> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>axis</strong> (<em>int</em><em>, </em><em>default -1</em>) – The axis that should be normalized. This is typically the axis of the channels.</li> |
| <li><strong>epsilon</strong> (<em>float</em><em>, </em><em>default 1e-5</em>) – Small float added to variance to avoid dividing by zero.</li> |
| <li><strong>center</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, add offset of <cite>beta</cite> to normalized tensor. |
| If False, <cite>beta</cite> is ignored.</li> |
| <li><strong>scale</strong> (<em>bool</em><em>, </em><em>default True</em>) – If True, multiply by <cite>gamma</cite>. If False, <cite>gamma</cite> is not used.</li> |
| <li><strong>beta_initializer</strong> (str or <cite>Initializer</cite>, default ‘zeros’) – Initializer for the beta weight.</li> |
| <li><strong>gamma_initializer</strong> (str or <cite>Initializer</cite>, default ‘ones’) – Initializer for the gamma weight.</li> |
| <li><strong>in_channels</strong> (<em>int</em><em>, </em><em>default 0</em>) – Number of channels (feature maps) in input data. If not specified, |
| initialization will be deferred to the first time <cite>forward</cite> is called |
| and <cite>in_channels</cite> will be inferred from the shape of input data.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">References</p> |
| <p><a class="reference external" href="https://arxiv.org/pdf/1607.06450.pdf">Layer Normalization</a></p> |
| <p class="rubric">Examples</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="c1"># Input of shape (2, 5)</span> |
| <span class="gp">>>> </span><span class="n">x</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">array</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">2</span><span class="p">]])</span> |
| <span class="gp">>>> </span><span class="c1"># Layer normalization is calculated with the above formula</span> |
| <span class="gp">>>> </span><span class="n">layer</span> <span class="o">=</span> <span class="n">LayerNorm</span><span class="p">()</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="o">.</span><span class="n">initialize</span><span class="p">(</span><span class="n">ctx</span><span class="o">=</span><span class="n">mx</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">layer</span><span class="p">(</span><span class="n">x</span><span class="p">)</span> |
| <span class="go">[[-1.41421 -0.707105 0. 0.707105 1.41421 ]</span> |
| <span class="go"> [-1.2247195 -1.2247195 0.81647956 0.81647956 0.81647956]]</span> |
| <span class="go"><NDArray 2x5 @cpu(0)></span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.LeakyReLU"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">LeakyReLU</code><span class="sig-paren">(</span><em>alpha</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#LeakyReLU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.LeakyReLU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Leaky version of a Rectified Linear Unit.</p> |
| <p>It allows a small gradient when the unit is not active</p> |
| <div class="math"> |
| \[\begin{split}f\left(x\right) = \left\{ |
| \begin{array}{lr} |
| \alpha x & : x \lt 0 \\ |
| x & : x \geq 0 \\ |
| \end{array} |
| \right.\\\end{split}\]</div> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><strong>alpha</strong> (<em>float</em>) – slope coefficient for the negative half axis. Must be >= 0.</td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.MaxPool1D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">MaxPool1D</code><span class="sig-paren">(</span><em>pool_size=2</em>, <em>strides=None</em>, <em>padding=0</em>, <em>layout='NCW'</em>, <em>ceil_mode=False</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#MaxPool1D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.MaxPool1D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Max pooling operation for one dimensional data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>pool_size</strong> (<em>int</em>) – Size of the max pooling windows.</li> |
| <li><strong>strides</strong> (<em>int</em><em>, or </em><em>None</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</li> |
| <li><strong>padding</strong> (<em>int</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCW'</em>) – Dimension ordering of data and weight. Can be ‘NCW’, ‘NWC’, etc. |
| ‘N’, ‘C’, ‘W’ stands for batch, channel, and width (time) dimensions |
| respectively. Pooling is applied on the W dimension.</li> |
| <li><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 3D input tensor with shape <cite>(batch_size, in_channels, width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 3D output tensor with shape <cite>(batch_size, channels, out_width)</cite> |
| when <cite>layout</cite> is <cite>NCW</cite>. out_width is calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="o">-</span><span class="n">pool_size</span><span class="p">)</span><span class="o">/</span><span class="n">strides</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.MaxPool2D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">MaxPool2D</code><span class="sig-paren">(</span><em>pool_size=(2</em>, <em>2)</em>, <em>strides=None</em>, <em>padding=0</em>, <em>layout='NCHW'</em>, <em>ceil_mode=False</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#MaxPool2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.MaxPool2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Max pooling operation for two dimensional (spatial) data.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – Size of the max pooling windows.</li> |
| <li><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 2 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 2 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCHW'</em>) – Dimension ordering of data and weight. Can be ‘NCHW’, ‘NHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’ stands for batch, channel, height, and width |
| dimensions respectively. padding is applied on ‘H’ and ‘W’ dimension.</li> |
| <li><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 4D input tensor with shape |
| <cite>(batch_size, in_channels, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 4D output tensor with shape |
| <cite>(batch_size, channels, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.MaxPool3D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">MaxPool3D</code><span class="sig-paren">(</span><em>pool_size=(2</em>, <em>2</em>, <em>2)</em>, <em>strides=None</em>, <em>padding=0</em>, <em>ceil_mode=False</em>, <em>layout='NCDHW'</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#MaxPool3D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.MaxPool3D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Max pooling operation for 3D data (spatial or spatio-temporal).</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first last simple"> |
| <li><strong>pool_size</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – Size of the max pooling windows.</li> |
| <li><strong>strides</strong> (<em>int</em><em>, </em><em>list/tuple of 3 ints</em><em>, or </em><em>None.</em>) – Factor by which to downscale. E.g. 2 will halve the input size. |
| If <cite>None</cite>, it will default to <cite>pool_size</cite>.</li> |
| <li><strong>padding</strong> (<em>int</em><em> or </em><em>list/tuple of 3 ints</em><em>,</em>) – If padding is non-zero, then the input is implicitly |
| zero-padded on both sides for padding number of points.</li> |
| <li><strong>layout</strong> (<em>str</em><em>, </em><em>default 'NCDHW'</em>) – Dimension ordering of data and weight. Can be ‘NCDHW’, ‘NDHWC’, etc. |
| ‘N’, ‘C’, ‘H’, ‘W’, ‘D’ stands for batch, channel, height, width and |
| depth dimensions respectively. padding is applied on ‘D’, ‘H’ and ‘W’ |
| dimension.</li> |
| <li><strong>ceil_mode</strong> (<em>bool</em><em>, </em><em>default False</em>) – When <cite>True</cite>, will use ceil instead of floor to compute the output shape.</li> |
| </ul> |
| </td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: 5D input tensor with shape |
| <cite>(batch_size, in_channels, depth, height, width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| For other layouts shape is permuted accordingly.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: 5D output tensor with shape |
| <cite>(batch_size, channels, out_depth, out_height, out_width)</cite> when <cite>layout</cite> is <cite>NCW</cite>. |
| out_depth, out_height and out_width are calculated as:</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">out_depth</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">depth</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_height</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">height</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| <span class="n">out_width</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">width</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">padding</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span><span class="o">-</span><span class="n">pool_size</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">/</span><span class="n">strides</span><span class="p">[</span><span class="mi">2</span><span class="p">])</span><span class="o">+</span><span class="mi">1</span> |
| </pre></div> |
| </div> |
| <p>When <cite>ceil_mode</cite> is <cite>True</cite>, ceil will be used instead of floor in this |
| equation.</p> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.PReLU"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">PReLU</code><span class="sig-paren">(</span><em>alpha_initializer=<mxnet.initializer.Constant object></em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#PReLU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.PReLU" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Parametric leaky version of a Rectified Linear Unit. |
| <<a class="reference external" href="https://arxiv.org/abs/1502.01852">https://arxiv.org/abs/1502.01852</a>>`_ paper.</p> |
| <p>It learns a gradient when the unit is not active</p> |
| <div class="math"> |
| \[\begin{split}f\left(x\right) = \left\{ |
| \begin{array}{lr} |
| \alpha x & : x \lt 0 \\ |
| x & : x \geq 0 \\ |
| \end{array} |
| \right.\\\end{split}\]</div> |
| <p>where alpha is a learned parameter.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><strong>alpha_initializer</strong> (<a class="reference internal" href="../optimization/optimization.html#mxnet.initializer.Initializer" title="mxnet.initializer.Initializer"><em>Initializer</em></a>) – Initializer for the <cite>embeddings</cite> matrix.</td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.ReflectionPad2D"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">ReflectionPad2D</code><span class="sig-paren">(</span><em>padding=0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/conv_layers.html#ReflectionPad2D"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.ReflectionPad2D" title="Permalink to this definition">¶</a></dt> |
| <dd><p>Pads the input tensor using the reflection of the input boundary.</p> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><strong>padding</strong> (<em>int</em>) – An integer padding size</td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with the shape <span class="math">\((N, C, H_{in}, W_{in})\)</span>.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last"> |
| <li><p class="first"><strong>out</strong>: output tensor with the shape <span class="math">\((N, C, H_{out}, W_{out})\)</span>, where</p> |
| <div class="math"> |
| \[ \begin{align}\begin{aligned}H_{out} = H_{in} + 2 \cdot padding\\W_{out} = W_{in} + 2 \cdot padding\end{aligned}\end{align} \]</div> |
| </li> |
| </ul> |
| </dd> |
| </dl> |
| <p class="rubric">Examples</p> |
| <div class="highlight-default"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">m</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">ReflectionPad2D</span><span class="p">(</span><span class="mi">3</span><span class="p">)</span> |
| <span class="gp">>>> </span><span class="nb">input</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">normal</span><span class="p">(</span><span class="n">shape</span><span class="o">=</span><span class="p">(</span><span class="mi">16</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">224</span><span class="p">,</span> <span class="mi">224</span><span class="p">))</span> |
| <span class="gp">>>> </span><span class="n">output</span> <span class="o">=</span> <span class="n">m</span><span class="p">(</span><span class="nb">input</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.SELU"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">SELU</code><span class="sig-paren">(</span><em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#SELU"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.SELU" title="Permalink to this definition">¶</a></dt> |
| <dd><dl class="docutils"> |
| <dt>Scaled Exponential Linear Unit (SELU)</dt> |
| <dd>“Self-Normalizing Neural Networks”, Klambauer et al, 2017 |
| <a class="reference external" href="https://arxiv.org/abs/1706.02515">https://arxiv.org/abs/1706.02515</a></dd> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <dl class="class"> |
| <dt id="mxnet.gluon.nn.Swish"> |
| <em class="property">class </em><code class="descclassname">mxnet.gluon.nn.</code><code class="descname">Swish</code><span class="sig-paren">(</span><em>beta=1.0</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="reference internal" href="../../../_modules/mxnet/gluon/nn/activations.html#Swish"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#mxnet.gluon.nn.Swish" title="Permalink to this definition">¶</a></dt> |
| <dd><dl class="docutils"> |
| <dt>Swish Activation function</dt> |
| <dd><a class="reference external" href="https://arxiv.org/pdf/1710.05941.pdf">https://arxiv.org/pdf/1710.05941.pdf</a></dd> |
| </dl> |
| <table class="docutils field-list" frame="void" rules="none"> |
| <col class="field-name"/> |
| <col class="field-body"/> |
| <tbody valign="top"> |
| <tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><strong>beta</strong> (<em>float</em>) – swish(x) = x * sigmoid(beta*x)</td> |
| </tr> |
| </tbody> |
| </table> |
| <dl class="docutils"> |
| <dt>Inputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>data</strong>: input tensor with arbitrary shape.</li> |
| </ul> |
| </dd> |
| <dt>Outputs:</dt> |
| <dd><ul class="first last simple"> |
| <li><strong>out</strong>: output tensor with the same shape as <cite>data</cite>.</li> |
| </ul> |
| </dd> |
| </dl> |
| </dd></dl> |
| <script>auto_index("api-reference");</script></div> |
| </div> |
| </div> |
| </div> |
| <div aria-label="main navigation" class="sphinxsidebar rightsidebar" role="navigation"> |
| <div class="sphinxsidebarwrapper"> |
| <h3><a href="../../../index.html">Table Of Contents</a></h3> |
| <ul> |
| <li><a class="reference internal" href="#">Gluon Neural Network Layers</a><ul> |
| <li><a class="reference internal" href="#overview">Overview</a></li> |
| <li><a class="reference internal" href="#basic-layers">Basic Layers</a></li> |
| <li><a class="reference internal" href="#convolutional-layers">Convolutional Layers</a></li> |
| <li><a class="reference internal" href="#pooling-layers">Pooling Layers</a></li> |
| <li><a class="reference internal" href="#activation-layers">Activation Layers</a></li> |
| <li><a class="reference internal" href="#api-reference">API Reference</a></li> |
| </ul> |
| </li> |
| </ul> |
| </div> |
| </div> |
| </div><div class="footer"> |
| <div class="section-disclaimer"> |
| <div class="container"> |
| <div> |
| <img height="60" src="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/apache_incubator_logo.png"/> |
| <p> |
| Apache MXNet is an effort undergoing incubation at The Apache Software Foundation (ASF), <strong>sponsored by the <i>Apache Incubator</i></strong>. Incubation is required of all newly accepted projects until a further review indicates that the infrastructure, communications, and decision making process have stabilized in a manner consistent with other successful ASF projects. While incubation status is not necessarily a reflection of the completeness or stability of the code, it does indicate that the project has yet to be fully endorsed by the ASF. |
| </p> |
| <p> |
| "Copyright © 2017-2018, The Apache Software Foundation |
| Apache MXNet, MXNet, Apache, the Apache feather, and the Apache MXNet project logo are either registered trademarks or trademarks of the Apache Software Foundation." |
| </p> |
| </div> |
| </div> |
| </div> |
| </div> <!-- pagename != index --> |
| </div> |
| <script crossorigin="anonymous" integrity="sha384-0mSbJDEHialfmuBBQP6A4Qrprq5OVfW37PRR3j5ELqxss1yVqOtnepnHVP9aJ7xS" src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.6/js/bootstrap.min.js"></script> |
| <script src="../../../_static/js/sidebar.js" type="text/javascript"></script> |
| <script src="../../../_static/js/search.js" type="text/javascript"></script> |
| <script src="../../../_static/js/navbar.js" type="text/javascript"></script> |
| <script src="../../../_static/js/clipboard.min.js" type="text/javascript"></script> |
| <script src="../../../_static/js/copycode.js" type="text/javascript"></script> |
| <script src="../../../_static/js/page.js" type="text/javascript"></script> |
| <script src="../../../_static/js/docversion.js" type="text/javascript"></script> |
| <script type="text/javascript"> |
| $('body').ready(function () { |
| $('body').css('visibility', 'visible'); |
| }); |
| </script> |
| </body> |
| </html> |
