Analytical programming with extended individuals

Abstract

This paper proposes a new technique for the estimation of values of constants in programs synthesized by Analytical Programming (AP). Proposed technique divides the features of an individual in Evolutionary Algorithm (EA) into two parts - program part and constant part. Features in program part are mapped to synthesized program and features in constant part are used as constant values. AP with implementation of this technique is tested on four benchmark functions - Quintic, Sextic, 3Sine and 4Sine. © ECMS Thorsten Claus, Frank Herrmann, Michael Manitz, Oliver Rose (Editors).