The Fitness Sharing Genetic Algorithms with Adaptive Power Law Scaling
应用自适应指数比例变换的适应值共享遗传算法

The fitness sharing genetic algorithms are the common approaches to solve multi-modal optimization problems. In this paper, a new adaptive power law scaling method is suggested to improve the search ability of the fitness sharing genetic algorithms. Different power law scaling methods have been adopted to optimize the massive deceptive problem. The empirical results show that the new adaptive power law scaling method can find all the global peaks steadily and quickly. This method is especially suit for the problems whose radii of peaks are hard to define.