Parallel multi-objective evolutionary algorithms on graphics processing units
Lingnan University, Tuen Mun, Hong Kong, Hong Kong
In GECCO ’09: Proceedings of the 11th Annual Conference Companion on Genetic and Evolutionary Computation Conference (2009), pp. 2515-2522.
Most real-life optimization problems or decision-making problems are multi-objective in nature, since they normally have several (possibly conflicting) objectives that must be satisfied at the same time. Multi-Objective Evolutionary Algorithms (MOEAs) have been gaining increasing attention among researchers and practitioners. However, they may execute for a long time for some difficult problems, because several evaluations must be performed. Moreover, the non-dominance checking and the non-dominated selection procedures are also very time consuming. From our experiments, more than 99% of the execution time is used in performing the two procedures. A promising approach to overcome this limitation is to parallelize these algorithms. In this paper, we propose a parallel MOEA on consumer-level Graphics Processing Units (GPU). We perform many experiments on two-objective and three-objective benchmark problems to compare our parallel MOEA with a sequential MOEA and demonstrate that the former is much more efficient than the latter.
November 22, 2010 by hgpu