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 your home. 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 the Genetic and Evolutionary Computation Conference 2016, GECCO 2016, pp. 773–780, ACM, Denver, Colorado, USA, 2016, ISBN: 978-1-4503-4206-3, (Best Paper Award in Genetic Programming).
• 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).