Incorporating linkage learning into the GeLog framework

Abstract

This article introduces modifications that have been applied to GeLog, a genetic logic programming framework, in order to improve its performance. The main emphasis of this work is the structure processing of genetic algorithms. As studies have shown, the linkage of genes plays an important role in the performance of genetic algorithms. Thus, different approaches that take linkage learning into account have been reviewed and the most promising has been implemented and tested with GeLog. It is demonstrated that the modified program solves problems that proved hard for the original system.