mirror of
https://github.com/gryf/coach.git
synced 2025-12-17 19:20:19 +01:00
618 lines
36 KiB
HTML
618 lines
36 KiB
HTML
|
||
|
||
<!DOCTYPE html>
|
||
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
|
||
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
|
||
<head>
|
||
<meta charset="utf-8">
|
||
|
||
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
||
|
||
<title>Exploration Policies — Reinforcement Learning Coach 0.12.0 documentation</title>
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
<script type="text/javascript" src="../../_static/js/modernizr.min.js"></script>
|
||
|
||
|
||
<script type="text/javascript" id="documentation_options" data-url_root="../../" src="../../_static/documentation_options.js"></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="../../_static/language_data.js"></script>
|
||
<script async="async" type="text/javascript" src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/latest.js?config=TeX-AMS-MML_HTMLorMML"></script>
|
||
|
||
<script type="text/javascript" src="../../_static/js/theme.js"></script>
|
||
|
||
|
||
|
||
|
||
<link rel="stylesheet" href="../../_static/css/theme.css" type="text/css" />
|
||
<link rel="stylesheet" href="../../_static/pygments.css" type="text/css" />
|
||
<link rel="stylesheet" href="../../_static/css/custom.css" type="text/css" />
|
||
<link rel="index" title="Index" href="../../genindex.html" />
|
||
<link rel="search" title="Search" href="../../search.html" />
|
||
<link rel="next" title="Filters" href="../filters/index.html" />
|
||
<link rel="prev" title="Environments" href="../environments/index.html" />
|
||
<link href="../../_static/css/custom.css" rel="stylesheet" type="text/css">
|
||
|
||
</head>
|
||
|
||
<body class="wy-body-for-nav">
|
||
|
||
|
||
<div class="wy-grid-for-nav">
|
||
|
||
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
|
||
<div class="wy-side-scroll">
|
||
<div class="wy-side-nav-search" >
|
||
|
||
|
||
|
||
<a href="../../index.html" class="icon icon-home"> Reinforcement Learning Coach
|
||
|
||
|
||
|
||
|
||
<img src="../../_static/dark_logo.png" class="logo" alt="Logo"/>
|
||
|
||
</a>
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
<div role="search">
|
||
<form id="rtd-search-form" class="wy-form" action="../../search.html" method="get">
|
||
<input type="text" name="q" placeholder="Search docs" />
|
||
<input type="hidden" name="check_keywords" value="yes" />
|
||
<input type="hidden" name="area" value="default" />
|
||
</form>
|
||
</div>
|
||
|
||
|
||
</div>
|
||
|
||
<div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation">
|
||
|
||
|
||
|
||
|
||
|
||
|
||
<p class="caption"><span class="caption-text">Intro</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../usage.html">Usage</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../dist_usage.html">Usage - Distributed Coach</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../features/index.html">Features</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../selecting_an_algorithm.html">Selecting an Algorithm</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../dashboard.html">Coach Dashboard</a></li>
|
||
</ul>
|
||
<p class="caption"><span class="caption-text">Design</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../design/control_flow.html">Control Flow</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../design/network.html">Network Design</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../design/horizontal_scaling.html">Distributed Coach - Horizontal Scale-Out</a></li>
|
||
</ul>
|
||
<p class="caption"><span class="caption-text">Contributing</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../contributing/add_agent.html">Adding a New Agent</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../../contributing/add_env.html">Adding a New Environment</a></li>
|
||
</ul>
|
||
<p class="caption"><span class="caption-text">Components</span></p>
|
||
<ul class="current">
|
||
<li class="toctree-l1"><a class="reference internal" href="../agents/index.html">Agents</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../architectures/index.html">Architectures</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../data_stores/index.html">Data Stores</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../environments/index.html">Environments</a></li>
|
||
<li class="toctree-l1 current"><a class="current reference internal" href="#">Exploration Policies</a><ul>
|
||
<li class="toctree-l2"><a class="reference internal" href="#explorationpolicy">ExplorationPolicy</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#additivenoise">AdditiveNoise</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#boltzmann">Boltzmann</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#bootstrapped">Bootstrapped</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#categorical">Categorical</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#continuousentropy">ContinuousEntropy</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#egreedy">EGreedy</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#greedy">Greedy</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#ouprocess">OUProcess</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#parameternoise">ParameterNoise</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#truncatednormal">TruncatedNormal</a></li>
|
||
<li class="toctree-l2"><a class="reference internal" href="#ucb">UCB</a></li>
|
||
</ul>
|
||
</li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../filters/index.html">Filters</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../memories/index.html">Memories</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../memory_backends/index.html">Memory Backends</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../orchestrators/index.html">Orchestrators</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../core_types.html">Core Types</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../spaces.html">Spaces</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="../additional_parameters.html">Additional Parameters</a></li>
|
||
</ul>
|
||
|
||
|
||
|
||
</div>
|
||
</div>
|
||
</nav>
|
||
|
||
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
|
||
|
||
|
||
<nav class="wy-nav-top" aria-label="top navigation">
|
||
|
||
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
|
||
<a href="../../index.html">Reinforcement Learning Coach</a>
|
||
|
||
</nav>
|
||
|
||
|
||
<div class="wy-nav-content">
|
||
|
||
<div class="rst-content">
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
|
||
<div role="navigation" aria-label="breadcrumbs navigation">
|
||
|
||
<ul class="wy-breadcrumbs">
|
||
|
||
<li><a href="../../index.html">Docs</a> »</li>
|
||
|
||
<li>Exploration Policies</li>
|
||
|
||
|
||
<li class="wy-breadcrumbs-aside">
|
||
|
||
|
||
<a href="../../_sources/components/exploration_policies/index.rst.txt" rel="nofollow"> View page source</a>
|
||
|
||
|
||
</li>
|
||
|
||
</ul>
|
||
|
||
|
||
<hr/>
|
||
</div>
|
||
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
|
||
<div itemprop="articleBody">
|
||
|
||
<div class="section" id="exploration-policies">
|
||
<h1>Exploration Policies<a class="headerlink" href="#exploration-policies" title="Permalink to this headline">¶</a></h1>
|
||
<p>Exploration policies are a component that allow the agent to tradeoff exploration and exploitation according to a
|
||
predefined policy. This is one of the most important aspects of reinforcement learning agents, and can require some
|
||
tuning to get it right. Coach supports several pre-defined exploration policies, and it can be easily extended with
|
||
custom policies. Note that not all exploration policies are expected to work for both discrete and continuous action
|
||
spaces.</p>
|
||
<table class="docutils align-center">
|
||
<colgroup>
|
||
<col style="width: 35%" />
|
||
<col style="width: 37%" />
|
||
<col style="width: 29%" />
|
||
</colgroup>
|
||
<thead>
|
||
<tr class="row-odd"><th class="head"><p>Exploration Policy</p></th>
|
||
<th class="head"><p>Discrete Action Space</p></th>
|
||
<th class="head"><p>Box Action Space</p></th>
|
||
</tr>
|
||
</thead>
|
||
<tbody>
|
||
<tr class="row-even"><td><p>AdditiveNoise</p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-odd"><td><p>Boltzmann</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
</tr>
|
||
<tr class="row-even"><td><p>Bootstrapped</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
</tr>
|
||
<tr class="row-odd"><td><p>Categorical</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
</tr>
|
||
<tr class="row-even"><td><p>ContinuousEntropy</p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-odd"><td><p>EGreedy</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-even"><td><p>Greedy</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-odd"><td><p>OUProcess</p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-even"><td><p>ParameterNoise</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-odd"><td><p>TruncatedNormal</p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
</tr>
|
||
<tr class="row-even"><td><p>UCB</p></td>
|
||
<td><p><span class="green">V</span></p></td>
|
||
<td><p><span class="red">X</span></p></td>
|
||
</tr>
|
||
</tbody>
|
||
</table>
|
||
<div class="section" id="explorationpolicy">
|
||
<h2>ExplorationPolicy<a class="headerlink" href="#explorationpolicy" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.exploration_policy.ExplorationPolicy">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.exploration_policy.</code><code class="descname">ExplorationPolicy</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/exploration_policy.html#ExplorationPolicy"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.exploration_policy.ExplorationPolicy" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>An exploration policy takes the predicted actions or action values from the agent, and selects the action to
|
||
actually apply to the environment using some predefined algorithm.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><p><strong>action_space</strong> – the action space used by the environment</p>
|
||
</dd>
|
||
</dl>
|
||
<dl class="method">
|
||
<dt id="rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.change_phase">
|
||
<code class="descname">change_phase</code><span class="sig-paren">(</span><em>phase</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/exploration_policy.html#ExplorationPolicy.change_phase"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.change_phase" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Change between running phases of the algorithm
|
||
:param phase: Either Heatup or Train
|
||
:return: none</p>
|
||
</dd></dl>
|
||
|
||
<dl class="method">
|
||
<dt id="rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.get_action">
|
||
<code class="descname">get_action</code><span class="sig-paren">(</span><em>action_values: List[Union[int, float, numpy.ndarray, List]]</em><span class="sig-paren">)</span> → Union[int, float, numpy.ndarray, List]<a class="reference internal" href="../../_modules/rl_coach/exploration_policies/exploration_policy.html#ExplorationPolicy.get_action"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.get_action" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Given a list of values corresponding to each action,
|
||
choose one actions according to the exploration policy
|
||
:param action_values: A list of action values
|
||
:return: The chosen action</p>
|
||
</dd></dl>
|
||
|
||
<dl class="method">
|
||
<dt id="rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.requires_action_values">
|
||
<code class="descname">requires_action_values</code><span class="sig-paren">(</span><span class="sig-paren">)</span> → bool<a class="reference internal" href="../../_modules/rl_coach/exploration_policies/exploration_policy.html#ExplorationPolicy.requires_action_values"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.requires_action_values" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Allows exploration policies to define if they require the action values for the current step.
|
||
This can save up a lot of computation. For example in e-greedy, if the random value generated is smaller
|
||
than epsilon, the action is completely random, and the action values don’t need to be calculated
|
||
:return: True if the action values are required. False otherwise</p>
|
||
</dd></dl>
|
||
|
||
<dl class="method">
|
||
<dt id="rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.reset">
|
||
<code class="descname">reset</code><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/exploration_policy.html#ExplorationPolicy.reset"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.exploration_policy.ExplorationPolicy.reset" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Used for resetting the exploration policy parameters when needed
|
||
:return: None</p>
|
||
</dd></dl>
|
||
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="additivenoise">
|
||
<h2>AdditiveNoise<a class="headerlink" href="#additivenoise" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.additive_noise.AdditiveNoise">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.additive_noise.</code><code class="descname">AdditiveNoise</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>noise_percentage_schedule: rl_coach.schedules.Schedule</em>, <em>evaluation_noise_percentage: float</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/additive_noise.html#AdditiveNoise"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.additive_noise.AdditiveNoise" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>AdditiveNoise is an exploration policy intended for continuous action spaces. It takes the action from the agent
|
||
and adds a Gaussian distributed noise to it. The amount of noise added to the action follows the noise amount that
|
||
can be given in two different ways:
|
||
1. Specified by the user as a noise schedule which is taken in percentiles out of the action space size
|
||
2. Specified by the agents action. In case the agents action is a list with 2 values, the 1st one is assumed to
|
||
be the mean of the action, and 2nd is assumed to be its standard deviation.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>noise_percentage_schedule</strong> – the schedule for the noise variance percentage relative to the absolute range
|
||
of the action space</p></li>
|
||
<li><p><strong>evaluation_noise_percentage</strong> – the noise variance percentage that will be used during evaluation phases</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="boltzmann">
|
||
<h2>Boltzmann<a class="headerlink" href="#boltzmann" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.boltzmann.Boltzmann">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.boltzmann.</code><code class="descname">Boltzmann</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>temperature_schedule: rl_coach.schedules.Schedule</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/boltzmann.html#Boltzmann"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.boltzmann.Boltzmann" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>The Boltzmann exploration policy is intended for discrete action spaces. It assumes that each of the possible
|
||
actions has some value assigned to it (such as the Q value), and uses a softmax function to convert these values
|
||
into a distribution over the actions. It then samples the action for playing out of the calculated distribution.
|
||
An additional temperature schedule can be given by the user, and will control the steepness of the softmax function.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>temperature_schedule</strong> – the schedule for the temperature parameter of the softmax</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="bootstrapped">
|
||
<h2>Bootstrapped<a class="headerlink" href="#bootstrapped" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.bootstrapped.Bootstrapped">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.bootstrapped.</code><code class="descname">Bootstrapped</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>epsilon_schedule: rl_coach.schedules.Schedule</em>, <em>evaluation_epsilon: float</em>, <em>architecture_num_q_heads: int</em>, <em>continuous_exploration_policy_parameters: rl_coach.exploration_policies.exploration_policy.ExplorationParameters = <rl_coach.exploration_policies.additive_noise.AdditiveNoiseParameters object></em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/bootstrapped.html#Bootstrapped"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.bootstrapped.Bootstrapped" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Bootstrapped exploration policy is currently only used for discrete action spaces along with the
|
||
Bootstrapped DQN agent. It assumes that there is an ensemble of network heads, where each one predicts the
|
||
values for all the possible actions. For each episode, a single head is selected to lead the agent, according
|
||
to its value predictions. In evaluation, the action is selected using a majority vote over all the heads
|
||
predictions.</p>
|
||
<div class="admonition note">
|
||
<p class="admonition-title">Note</p>
|
||
<p>This exploration policy will only work for Discrete action spaces with Bootstrapped DQN style agents,
|
||
since it requires the agent to have a network with multiple heads.</p>
|
||
</div>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>epsilon_schedule</strong> – a schedule for the epsilon values</p></li>
|
||
<li><p><strong>evaluation_epsilon</strong> – the epsilon value to use for evaluation phases</p></li>
|
||
<li><p><strong>continuous_exploration_policy_parameters</strong> – the parameters of the continuous exploration policy to use
|
||
if the e-greedy is used for a continuous policy</p></li>
|
||
<li><p><strong>architecture_num_q_heads</strong> – the number of q heads to select from</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="categorical">
|
||
<h2>Categorical<a class="headerlink" href="#categorical" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.categorical.Categorical">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.categorical.</code><code class="descname">Categorical</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/categorical.html#Categorical"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.categorical.Categorical" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Categorical exploration policy is intended for discrete action spaces. It expects the action values to
|
||
represent a probability distribution over the action, from which a single action will be sampled.
|
||
In evaluation, the action that has the highest probability will be selected. This is particularly useful for
|
||
actor-critic schemes, where the actors output is a probability distribution over the actions.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><p><strong>action_space</strong> – the action space used by the environment</p>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="continuousentropy">
|
||
<h2>ContinuousEntropy<a class="headerlink" href="#continuousentropy" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.continuous_entropy.ContinuousEntropy">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.continuous_entropy.</code><code class="descname">ContinuousEntropy</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>noise_percentage_schedule: rl_coach.schedules.Schedule</em>, <em>evaluation_noise_percentage: float</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/continuous_entropy.html#ContinuousEntropy"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.continuous_entropy.ContinuousEntropy" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>Continuous entropy is an exploration policy that is actually implemented as part of the network.
|
||
The exploration policy class is only a placeholder for choosing this policy. The exploration policy is
|
||
implemented by adding a regularization factor to the network loss, which regularizes the entropy of the action.
|
||
This exploration policy is only intended for continuous action spaces, and assumes that the entire calculation
|
||
is implemented as part of the head.</p>
|
||
<div class="admonition warning">
|
||
<p class="admonition-title">Warning</p>
|
||
<p>This exploration policy expects the agent or the network to implement the exploration functionality.
|
||
There are only a few heads that actually are relevant and implement the entropy regularization factor.</p>
|
||
</div>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>noise_percentage_schedule</strong> – the schedule for the noise variance percentage relative to the absolute range
|
||
of the action space</p></li>
|
||
<li><p><strong>evaluation_noise_percentage</strong> – the noise variance percentage that will be used during evaluation phases</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="egreedy">
|
||
<h2>EGreedy<a class="headerlink" href="#egreedy" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.e_greedy.EGreedy">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.e_greedy.</code><code class="descname">EGreedy</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>epsilon_schedule: rl_coach.schedules.Schedule</em>, <em>evaluation_epsilon: float</em>, <em>continuous_exploration_policy_parameters: rl_coach.exploration_policies.exploration_policy.ExplorationParameters = <rl_coach.exploration_policies.additive_noise.AdditiveNoiseParameters object></em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/e_greedy.html#EGreedy"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.e_greedy.EGreedy" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>e-greedy is an exploration policy that is intended for both discrete and continuous action spaces.</p>
|
||
<p>For discrete action spaces, it assumes that each action is assigned a value, and it selects the action with the
|
||
highest value with probability 1 - epsilon. Otherwise, it selects a action sampled uniformly out of all the
|
||
possible actions. The epsilon value is given by the user and can be given as a schedule.
|
||
In evaluation, a different epsilon value can be specified.</p>
|
||
<p>For continuous action spaces, it assumes that the mean action is given by the agent. With probability epsilon,
|
||
it samples a random action out of the action space bounds. Otherwise, it selects the action according to a
|
||
given continuous exploration policy, which is set to AdditiveNoise by default. In evaluation, the action is
|
||
always selected according to the given continuous exploration policy (where its phase is set to evaluation as well).</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>epsilon_schedule</strong> – a schedule for the epsilon values</p></li>
|
||
<li><p><strong>evaluation_epsilon</strong> – the epsilon value to use for evaluation phases</p></li>
|
||
<li><p><strong>continuous_exploration_policy_parameters</strong> – the parameters of the continuous exploration policy to use
|
||
if the e-greedy is used for a continuous policy</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="greedy">
|
||
<h2>Greedy<a class="headerlink" href="#greedy" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.greedy.Greedy">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.greedy.</code><code class="descname">Greedy</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/greedy.html#Greedy"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.greedy.Greedy" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>The Greedy exploration policy is intended for both discrete and continuous action spaces.
|
||
For discrete action spaces, it always selects the action with the maximum value, as given by the agent.
|
||
For continuous action spaces, it always return the exact action, as it was given by the agent.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><p><strong>action_space</strong> – the action space used by the environment</p>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="ouprocess">
|
||
<h2>OUProcess<a class="headerlink" href="#ouprocess" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.ou_process.OUProcess">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.ou_process.</code><code class="descname">OUProcess</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>mu: float = 0</em>, <em>theta: float = 0.15</em>, <em>sigma: float = 0.2</em>, <em>dt: float = 0.01</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/ou_process.html#OUProcess"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.ou_process.OUProcess" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>OUProcess exploration policy is intended for continuous action spaces, and selects the action according to
|
||
an Ornstein-Uhlenbeck process. The Ornstein-Uhlenbeck process implements the action as a Gaussian process, where
|
||
the samples are correlated between consequent time steps.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><p><strong>action_space</strong> – the action space used by the environment</p>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="parameternoise">
|
||
<h2>ParameterNoise<a class="headerlink" href="#parameternoise" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.parameter_noise.ParameterNoise">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.parameter_noise.</code><code class="descname">ParameterNoise</code><span class="sig-paren">(</span><em>network_params: Dict[str, rl_coach.base_parameters.NetworkParameters], action_space: rl_coach.spaces.ActionSpace</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/parameter_noise.html#ParameterNoise"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.parameter_noise.ParameterNoise" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>The ParameterNoise exploration policy is intended for both discrete and continuous action spaces.
|
||
It applies the exploration policy by replacing all the dense network layers with noisy layers.
|
||
The noisy layers have both weight means and weight standard deviations, and for each forward pass of the network
|
||
the weights are sampled from a normal distribution that follows the learned weights mean and standard deviation
|
||
values.</p>
|
||
<p>Warning: currently supported only by DQN variants</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><p><strong>action_space</strong> – the action space used by the environment</p>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="truncatednormal">
|
||
<h2>TruncatedNormal<a class="headerlink" href="#truncatednormal" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.truncated_normal.TruncatedNormal">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.truncated_normal.</code><code class="descname">TruncatedNormal</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>noise_percentage_schedule: rl_coach.schedules.Schedule</em>, <em>evaluation_noise_percentage: float</em>, <em>clip_low: float</em>, <em>clip_high: float</em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/truncated_normal.html#TruncatedNormal"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.truncated_normal.TruncatedNormal" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>The TruncatedNormal exploration policy is intended for continuous action spaces. It samples the action from a
|
||
normal distribution, where the mean action is given by the agent, and the standard deviation can be given in t
|
||
wo different ways:
|
||
1. Specified by the user as a noise schedule which is taken in percentiles out of the action space size
|
||
2. Specified by the agents action. In case the agents action is a list with 2 values, the 1st one is assumed to
|
||
be the mean of the action, and 2nd is assumed to be its standard deviation.
|
||
When the sampled action is outside of the action bounds given by the user, it is sampled again and again, until it
|
||
is within the bounds.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>noise_percentage_schedule</strong> – the schedule for the noise variance percentage relative to the absolute range
|
||
of the action space</p></li>
|
||
<li><p><strong>evaluation_noise_percentage</strong> – the noise variance percentage that will be used during evaluation phases</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
<div class="section" id="ucb">
|
||
<h2>UCB<a class="headerlink" href="#ucb" title="Permalink to this headline">¶</a></h2>
|
||
<dl class="class">
|
||
<dt id="rl_coach.exploration_policies.ucb.UCB">
|
||
<em class="property">class </em><code class="descclassname">rl_coach.exploration_policies.ucb.</code><code class="descname">UCB</code><span class="sig-paren">(</span><em>action_space: rl_coach.spaces.ActionSpace</em>, <em>epsilon_schedule: rl_coach.schedules.Schedule</em>, <em>evaluation_epsilon: float</em>, <em>architecture_num_q_heads: int</em>, <em>lamb: int</em>, <em>continuous_exploration_policy_parameters: rl_coach.exploration_policies.exploration_policy.ExplorationParameters = <rl_coach.exploration_policies.additive_noise.AdditiveNoiseParameters object></em><span class="sig-paren">)</span><a class="reference internal" href="../../_modules/rl_coach/exploration_policies/ucb.html#UCB"><span class="viewcode-link">[source]</span></a><a class="headerlink" href="#rl_coach.exploration_policies.ucb.UCB" title="Permalink to this definition">¶</a></dt>
|
||
<dd><p>UCB exploration policy is following the upper confidence bound heuristic to sample actions in discrete action spaces.
|
||
It assumes that there are multiple network heads that are predicting action values, and that the standard deviation
|
||
between the heads predictions represents the uncertainty of the agent in each of the actions.
|
||
It then updates the action value estimates to by mean(actions)+lambda*stdev(actions), where lambda is
|
||
given by the user. This exploration policy aims to take advantage of the uncertainty of the agent in its predictions,
|
||
and select the action according to the tradeoff between how uncertain the agent is, and how large it predicts
|
||
the outcome from those actions to be.</p>
|
||
<dl class="field-list simple">
|
||
<dt class="field-odd">Parameters</dt>
|
||
<dd class="field-odd"><ul class="simple">
|
||
<li><p><strong>action_space</strong> – the action space used by the environment</p></li>
|
||
<li><p><strong>epsilon_schedule</strong> – a schedule for the epsilon values</p></li>
|
||
<li><p><strong>evaluation_epsilon</strong> – the epsilon value to use for evaluation phases</p></li>
|
||
<li><p><strong>architecture_num_q_heads</strong> – the number of q heads to select from</p></li>
|
||
<li><p><strong>lamb</strong> – lambda coefficient for taking the standard deviation into account</p></li>
|
||
<li><p><strong>continuous_exploration_policy_parameters</strong> – the parameters of the continuous exploration policy to use
|
||
if the e-greedy is used for a continuous policy</p></li>
|
||
</ul>
|
||
</dd>
|
||
</dl>
|
||
</dd></dl>
|
||
|
||
</div>
|
||
</div>
|
||
|
||
|
||
</div>
|
||
|
||
</div>
|
||
<footer>
|
||
|
||
<div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
|
||
|
||
<a href="../filters/index.html" class="btn btn-neutral float-right" title="Filters" accesskey="n" rel="next">Next <span class="fa fa-arrow-circle-right"></span></a>
|
||
|
||
|
||
<a href="../environments/index.html" class="btn btn-neutral float-left" title="Environments" accesskey="p" rel="prev"><span class="fa fa-arrow-circle-left"></span> Previous</a>
|
||
|
||
</div>
|
||
|
||
|
||
<hr/>
|
||
|
||
<div role="contentinfo">
|
||
<p>
|
||
© Copyright 2018-2019, Intel AI Lab
|
||
|
||
</p>
|
||
</div>
|
||
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
|
||
|
||
</footer>
|
||
|
||
</div>
|
||
</div>
|
||
|
||
</section>
|
||
|
||
</div>
|
||
|
||
|
||
|
||
<script type="text/javascript">
|
||
jQuery(function () {
|
||
SphinxRtdTheme.Navigation.enable(true);
|
||
});
|
||
</script>
|
||
|
||
|
||
|
||
|
||
|
||
|
||
</body>
|
||
</html> |