The use of multiploid structures for individuals in evolutionary algorithms has been shown to have the advantage of including redundant information, increasing population diversity and in some cases improving non-stationary function optimisation performance. These advantages can translate into improved avoidance of premature convergence and an ability to cope with complex problems. However, as multiple information for the same trait is available, a method of gene selection or activation is required. This paper describes a dynamic decision method for gene selection, presents proof of concept results for this type of structure and outlines proposed benefits and applications.
Lecture notes in computer science: Lecture notes in artificial intelligence: Engineering of intelligent systems: Proceedings of the 14th International Conference on Industrial and Engineering Applications of Artificial Intelligence and Expert Systems (IEA/AIE 2001), Budapest, Hungary, 04-07 June 2001 / Laszlo Monostori, Jozsef Vancza and Moonis Ali (eds.),
Vol. 2070, pp. 374-382