Searching under multievolutionary pressures.pdf

Searching Under Multi-Evolutionary Pressures Hussein A. Abbass and Kalyanmoy Deb Artificial Life and Adaptive Robotics (A.L.A.R.) Lab, School of Computer Science, University of New South Wales, Australian Defence Force Academy Campus, Canberra, Australia. h.abbass@adfa.edu.au Mechanical Engineering Department, Indian Institute of Technology, Kanpur, Kanpur, PIN 208 016, India. deb@iitk.ac.in Abstract. A number of authors made the claim that a multiobjective approach preserves genetic diversity better than a single objective ap- proach. Sofar, none of these claims presented a thorough analysis to the effect of multiobjective approaches. In this paper, we provide such anal- ysis and show that a multiobjective approach does preserve reproductive diversity. We make our case by comparing a pareto multiobjective ap- proach against a single objective approach for solving single objective global optimization problems in the absence of mutation. We show that the fitness landscape is different in both cases and the multiobjective approach scales faster and produces better solutions than the single ob- jective approach. Keywords: artificial evolution; diversity; multiobjective optimization. 1 Introduction In biological systems, natural selection is carried out on multi-traits that are usually in conflict. In real life animal breeding, a farmer selects those animals to cull or mate based on a multi-trait selection index. Yet, in applying artificial evolutionary systems to solve real life problems, researchers deviate from the biological phenomena claiming that a biological characteristic may not be the idle choice for a computational problem. Although this assertion contains some truth, it