DanielKneipp / gsgp-gd

Geometric semantic genetic programming with geometric dispersion operators

gsgp-gd

Geometric semantic genetic programming with geometric dispersion operators

Description

This project is a Java implementation of the Geometric Semantic Genetic Programming. It also includes the geometric dispersion operator, presented in the papers below.

Instructions

Directory structure:

• src/edu/gsgp: source code.
• nbproject: project files generated by NetBeans.
• experiments:
  • data: datasets used in the experiments.
  • scripts: parameter files.
• dist: The compiled version (jar).

Usage

The parameter files are adjusted to work from the home directory. Thus, the folder experiments has to be copied to ~/.gsgp-gd/. In order to execute the experiments, run the following command line:

	java -jar GSGP.jar -p path/to/parameter/file

with one of the parameter files inside the folders bestGD (for the experiments with GSGP+GD) and gsgp (for experiments without the GD operator).

Parameters

The parameter files inside the directory experiments/scripts presents instructions for each parameter (presented as comments started with '#')

Published papers

• Luiz O. V. B. Oliveira and Fernando E. B. Otero and Gisele L. Pappa. A Dispersion Operator for Geometric Semantic Genetic Programming. Proc. of GECCO'16. 2016 (to appear).
• Luiz Otavio V. B. Oliveira and Fernando E. B. Otero and Gisele L. Pappa. A Generic Framework for Building Dispersion Operators in the Semantic Space. Genetic Programming Theory and Practice XIV. 2016 (to appear).