A New Dimension Division Scheme for Heterogeneous Multi-Population Cultural Algorithm
Ramin Raeesi, Joshua Chittle, Ziad Kobti · The Florida AI Research Society · 2014
Heterogeneous Multi-Population Cultural Algorithm (HMP-CA) is a new class of Multi-Population Cultural Algorithms which incorporates a number of local Cultural Algorithms (CAs) designed to optimize different subsets of the dimensions of a given problem. In this article, various dimension decomposition techniques for HMP-CAs are proposed and compared. The concept of using a dimension decomposition scheme which does not result in populations having the same number of dimensions is implemented, and named imbalanced dimension division. All the techniques are evaluated using a number of well-known benchmark optimization functions and two measures are defined in order to compare them including success rate and convergence ratio. The results show that imbalanced dimension division schema works better with a higher number of local CAs, and outperforms all the other evaluated techniques in both measures.