Genetic Programming with increased locality for Bloat control

Published in Projects
Project details
 Description
ID TIJ-ING-2012-110
Duration 2012-2014
Funding Dirección General de Educación Superior Tecnológica – México
Objective Develop a new Genetic Programming algorithm that minimizes the effects of Bloat and increases the representations locality during the search
Collaborations

 

Summary While GP has proven itself to be a powerful and flexible problem solving tool, it still suffers from several shortcomings that limits its use and acceptance in some domains. In general, the most important drawback of GP is its inefficiencies, caused by the Bloat phenomenon and its reliance on representations with a low locality that make the search operators less effective.
In this work, we are proposing to abandon the focus of the traditional search based on fitness and semantics, and are instead focusing on program behavior and solution novelty
Feedback