📄 starttofinish.htm
字号:
<html>
<head>
<meta name="GENERATOR" content="Microsoft FrontPage 5.0">
<meta name="ProgId" content="FrontPage.Editor.Document">
<meta http-equiv="Content-Type" content="text/html; charset=windows-1252">
<title>Genetic Programming FAQ</title>
</head>
<body>
<div id="nsbanner">
<table cellSpacing="0" cellPadding="0" width="100%" border="0">
<tr>
<td>
<div id="bannerrow1">
<table class="bannerparthead" cellSpacing="0">
<tr id="hdr">
<td class="runninghead" style="font-size: 70%">
<font face="Arial"><a class="show" id="design_toggle" title="Show or hide design notes" href="index.html"><img src="Images/Icons/G.png" width="14" height="14" border="0" align="middle"/></a> Genetic Programming Engine FAQ</font></td>
</tr>
</table>
</div>
<div id="TitleRow">
<h1 class="dtH1" style="font-size: 85%; left: 5px; position: relative">
<font face="Arial">Start to Finish</font></h1>
</div>
</td>
<td vAlign="top" align="right">
<font face="Arial">
<img src="Images/Icons/GP_blue.png" width="191" height="39"></font></td>
</tr>
</table>
</div>
<div id="nstext">
<h4 class="dtH4"><font face="Arial"><span style="font-weight: 400">To get a
basic understanding of the </span>GPEngine<span style="font-weight: 400">, it
is possible to run the default problem (the Artificial Ant) without any
knowledge of Genetic Programming. No </span>EngineOptions<span style="font-weight: 400">
or any other settings need to be set to experiment with the engine. The
first step is to run either the</span> EngineGUI<span style="font-weight: 400">.exe
(for a graphical user interface) or the </span>EngineConsole<span style="font-weight: 400">.exe
(for a command line interface) program. Upon execution, the GUI version
will load the default problem. All that is needed to begin is to click
the run button. The GUI will display statistics and individual history
dynamically. The default problem will place output into the directory
called AntIndividual. Running the console version requires no further
action. The code files will be placed into a AntIndividual directory.
The engine will run through the default of 100 generations each with a </span>
Population<span style="font-weight: 400"> of 100 individuals. The engine
will use a process similar to evolution's mutation and natural selection
(i.e., the top 10% of individuals pass "genes" to the next generation).
Fitness should improve over time and can be seen in the graphs in the GUI or
by seeing fitness values increasing in the command line version. Due to
the randomness of selection, the optimal solution may not be found before
reaching the generation limit. The generation limit can be increased or
the Population "reset" to start with a new Population of random individuals.
</span></font></h4>
<p class="dtH4"><font face="Arial">Once the basic working of the GPEngine is
understood, it is possible to modify EngineOptions to vary the performance of
the engine. Factors such as Population size and mutation rate can be
modified in order to try to find better individuals. Different
individuals can be selected. For example, selecting the GapAntIndividual
allows the individual ants to see gaps in the trail (as opposed to an
AntIndividual who cannot see the gaps). This typically results in fitter
individuals and a solution being reached faster. Different <b>
IEnvironment</b> initialization files may also be selected in order to let the
individual ants be tested with different trails. </font></p>
<p class="dtH4"><font face="Arial">It is also possible to create define new
problems to be solved by the GPEngine. Once a problem has been
identified, the Traveling Salesperson Problem (TSP) for example, it needs to
be analyzed and <b>translated into computer code</b>. A minimum of three
classes should be created to define the new individual and its environment.
The IIndividual, the IEnvironment, and the IGraphicHistory classes also known
as the problem space classes. Different IIndividuals can be defined to
represent the salesperson. For example, using the TSP, there can be
differences in how the next city to visit is chosen. All information the
computer needs to represent the salesperson's actions would need to be defined
such as moving from city to city. The environment would be defined
specifying the locations of the cities to visit and any limits the individual
has in the environment such as step limit. The IGraphicHistory would be
defined to track the fitness of the individual (i.e., shortest path to visit
all cities) as well as any methods to render a graphical representation of the
individuals path. </font></p>
<p class="dtH4"><font face="Arial">Finally, it is also possible to modify the
internal structure of the GPEngine. There are no limits to what can be
changed in the engine. For example, different selection and mutation
methods can be defined to determine how the next population is created.
</font></p>
<h4 class="dtH4"> </h4>
<h4 class="dtH4"><font face="Arial">See Also</font></h4>
<p align="left">
<font face="Arial"><a href="Console.htm">Console</a> | <a href="GeneticProgrammingEngine/GPE.htm">GPEngine</a> |
<a href="GeneticProgrammingEngine/Internal/EngineOptions.htm">EngineOptions</a>
| <a href="GUI.htm">GUI</a> |
<a href="GeneticProgrammingEngine/ProblemSpace/Environment.htm">IEnvironment</a>
| <a href="GeneticProgrammingEngine/ProblemSpace/Individual.htm">IIndividual</a> |
<a href="GeneticProgrammingEngine/Internal/Population.htm">Population</a> |
<a href="translate.htm">Translating a Problem Into Computer Code</a></font></p>
<hr>
<div id="footer">
<font face="Arial">
<a href="http://www.opensource.org/" target="_top">
<img style="padding-bottom: 5px" src="http://opensource.org/trademarks/open_source_button.png" align="left" border="0"></a>
This software licensed under the
<a href="http://opensource.org/licenses/afl-2.1.php" target="_top">Academic
Free License version 2.1</a>. (<a href="afl-2.1.txt">Plain
text</a>)</font></div>
</div>
</body>
</html>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -