Maximum a Posteriori Based Evolutionary Algorithm
暂无分享,去创建一个
This work is dedicated to the presentation and the analysis of the performance of Maximum a posteriori based Evolutionary Algorithm (MEA). MEA allows a hybridization of set of operators to preserve the diversity during the search. This approach is based on a set of search strategies which are composed of one crossover with one mutation method, respectively. The algorithm uses the Maximum a Posteriori Principle (MAP) to select the most probable strategy from those available in the search set. Experiments were performed on well-known continuous optimization problems to observe the impact of population size and operators rates on MEA’s behaviour, robustness and performance.
[1] Shigeyoshi Tsutsui,et al. Forking Genetic Algorithms: GAs with Search Space Division Schemes , 1997, Evolutionary Computation.
[2] Mark Wineberg,et al. The Shifting Balance Genetic Algorithm: improving the GA in a dynamic environment , 1999 .
[3] H. Shimodaira. DCGA: a diversity control oriented genetic algorithm , 1997 .
[4] Rasmus K. Ursem,et al. Diversity-Guided Evolutionary Algorithms , 2002, PPSN.