Analysis for Performance of Symbiosis Co-evolutionary Algorithm

In this paper, we analyze the behavior of symbiotic evolution algorithm for the N-Queens problem as benchmark problem for search methods in the field of aritificial intelligence. It is shown that this algorithm improves the ability of evolutionary search method. When the problem is solved by Genetic Algorithms (GAs), an ordinal representation is often used as one of gene conversion methods which convert from phenotype to genotype and reconvert. The representation can hinder occurrence of lethal genes. Typically, the representation pattern is fixed to one pattern. However, we consider that the kinds of generated solution are increased by preparing the permutation pattern with the several and the better solutions may be generated by the permutation pattern evolving. This paper introduces the symbiotic evolution model in which two evolutional populations are solutions and permutations for solving the N-Queens problem. To investigate the performance of symbiotic algorithm, it is compared with three kinds of evolutionary methods based on GA. From the results of computer simulations, we consider the characteristics of this algorithm