A GP neutral function for the artificial ANT problem

This paper introduces a function that increases the amount of neutrality (inactive code in Genetic Programming) for the Artificial Ant Problem. The objective of this approach is to try to smooth the ridged fitness landscape of the Santa Fe trail. Several experiments were carried out with different crossover and mutation rates, in order to identify the better settings to solve this problem and to compare the normal representation and the one proposed in this paper. The results indicate that the proposed approach is better than the conventional one. Also the difference between per individual and per node mutation is showed and a way to relate them is pointed out. Categories and Subject Descriptors I.2 [Artificial Intelligence]: Automatic Programming; D.2.8 [Software Engineering]: Metrics - complexity measures, performance measures General Terms Algorithms Keywords Evolutionary Computation, Genetic Programming, Neutrality

• Type: Conference
• Year: 2007
• Where: GECCO
• Authors: Esteban Ricalde, Katya Rodríguez-Vázquez
• Keywords: Artificial Ant Problem, GECCO 2007, Genetic Programming, Optimization, Ridged Fitness Landscape