Free cookie consent management tool by TermsFeed Policy Generator

Changes between Version 4 and Version 5 of Features/Genetic Programming


Ignore:
Timestamp:
01/02/12 16:05:08 (13 years ago)
Author:
gkronber
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • Features/Genetic Programming

    v4 v5  
    8484= Advanced Topics =
    8585
     86== Exporting Solutions ==
     87Solutions for symbolic regression or classification problems can be exported in different formats to make it possible to use the discovered formulas in other software. Currently we only support three textual export formats for data analysis models: MATLAB, LaTeX, and a generic prefix symbolic expression tree notation. However, it is very easy to implement you own formatter if necessary.
     88
     89The export functionality can be accessed via the `ViewHost` icon of the model view as shown in the screenshots below.
     90 1. Select the model of the discovered solution. The default view for the model shows the tree in a graphical form.
     91 2. By right-clicking the `ViewHost` icon the view displaying the model can be switched from the `Graphical Representation` to the `Textual Representation`
     92 3. Select the desired export format in the drop-down box.
     93
     94[[Image(http://dev.heuristiclab.com/trac/hl/core/raw-attachment/wiki/Genetic%20Programming/model_export1.png, width=300)]]
     95
     96[[Image(http://dev.heuristiclab.com/trac/hl/core/raw-attachment/wiki/Genetic%20Programming/model_export2.png, width=300)]]
     97
     98[[Image(http://dev.heuristiclab.com/trac/hl/core/raw-attachment/wiki/Genetic%20Programming/model_export3.png, width=300)]]
     99
     100
    86101== Automatically Defined Functions (ADF) ==
    87102We also provide an implementation for automatically defined functions (ADFs) as described by Koza. Automatically defined functions allow to extract useful code fragments into a separate function which can be called multiple times either while executing the result producing branch or another ADF. ADFs are created and manipulated by architecture altering operators while evolving genetic programming solutions. This means ADFs underlie selection pressure in the same way as the main result producing branch and the evolutionary process determines if ADFs survive in the final population and which code fragments are moved to ADFs.