Transposition versus Crossover: An Empirical Study Anabela Borges Simões Centre for Informatics and Systems of the University of Coimbra Polo II - Pinhal de Marrocos 3030 Coimbra - Portugal Ernesto Costa Centre for Informatics and Systems of the University of Coimbra Polo II - Pinhal de Marrocos 3030 Coimbra - Portugal Abstract Genetic algorithms are adaptive systems biologically motivated which have been used to solve different problems. Since Holland's proposals back in 1975, two main genetic operators, crossover and mutation, have been explored with success. Nonetheless, nature presents many other mechanisms of genetic recombination, based on phenomena like gene insertion, duplication or movement. The aim of this paper is to study one of these mechanisms: transposition. Transposition is a context-sensitive operator that promotes gene movement intra or inter chromosomes. This work presents an empirical study of the genetic algorithm performance, being the traditional crossover operator replaced by transposition. Such empirical study, based on an extensive set of test functions, shows that, under certain circumstances, transposition allows the GA to achieve higher quality solutions. 1 INTRODUCTION Genetic Algorithms (GA) are a search paradigm that applies ideas from evolutionary biology (crossover, mutation, natural selection) in order to deal with intractable search spaces (Holland 1992). The power and success of GA is mostly due to the diversity of the individuals of a population that evolve according to the principle of "the survival of the fittest". In the standard GA, the population diversity is obtained and maintained using the genetic operators of crossover and mutation, which allow the GA to find more promising solutions and avoid premature convergence to a local maximum (Goldberg 1989). In order to find the most efficient ways of using GA, many researchers have carried out extensive studies to understand specific aspects such as the role of types of selection, representation issues and how to apply different types of genetic operators. The use of the genetic operators has been the object of study of many researchers. Some important work related with crossover and mutation can be found in (Davis 1989; De Jong et al. 1992; Schaffer et al. 1991; Spears et al. 1991; Spears 1992; Spears 1993; Syswerda 1989). In addition to the traditional genetic operators, many authors have presented new genetic operators dependent of the problem domain, for instance, (Davidor 1989; D'Haeseleer 1993; Mathias et al. 1992; Parsons et al. 1995). Nevertheless, no new biologically inspired genetic operators have been widely adopted since the advent of GAs. Rather, the inversion operator included in John Holland's original work (Holland 1992) has been largely abandoned. Mitchell et al. (1994) point out the importance of studying new genetic operators. In addition, the authors emphasize the last discoveries of molecular biology as a good source of inspiration for new mechanisms of genetic material. Mitchell et al. (1994) and Mitchell (1996) state that it would be interesting to analyze if any of these mechanisms, incorporated in a GA, could lead to any significant advantages. Banzhaf et al. (1998) share the same opinion: the authors highlight the significance of implementing evolutionary approaches using mechanisms such as conjugation, transduction or transposition. Following these ideas, some authors have proposed other biologically inspired genetic operators, besides crossover and mutation. Furuhashi et al. (1994) introduced an application using a bacterial mechanism called transduction. Transduction is a process involving bacteriophages which carry a copy of a gene from a host cell and insert it in the chromosome of an infected cell. By transduction it is possible to spread the characteristics of a single bacterium to the rest of population. Furuhashi et al. (1994) presented a new approach for finding fuzzy rules for an obstacle avoidance problem involving a In Banzhaf, W., Daida, J., Eiben, A. E., Garzon, M. H., Honavar, V., Jakiela, M., and Smith, R. E. (eds.), Proceedings of the Genetic and Evolutionary Computation Conference (GECCO'99), Orlando, Florida USA, CA: Morgan Kaufmann.