A general framework of multi-population methods with clustering in undetectable dynamic environments.

Date
2012
Authors
Yang, Shengxiang
Journal Title
Journal ISSN
ISSN
1089-778X
Volume Title
Publisher
IEEE
Peer reviewed
Yes
Abstract
To solve dynamic optimization problems, multiple population methods are used to enhance the population diversity for an algorithm with the aim of maintaining multiple populations in different subareas in the fitness landscape. Many experimental studies have shown that locating and tracking multiple relatively good optima rather than a single global optimum is an effective idea in dynamic environments. However, several challenges need to be addressed when multipopulation methods are applied, e.g., how to create multiple populations, how to maintain them in different subareas, and how to deal with the situation where changes cannot be detected or predicted. To address these issues, this paper investigates a hierarchical clustering method to locate and track multiple optima for dynamic optimization problems. To deal with undetectable dynamic environments, this paper applies the random immigrants method without change detection based on a mechanism that can automatically reduce redundant individuals in the search space throughout the run. These methods are implemented into several research areas, including particle swarm optimization, genetic algorithm, and differential evolution. An experimental study is conducted based on the moving peaks benchmark to test the performance with several other algorithms from the literature. The experimental results show the efficiency of the clustering method for locating and tracking multiple optima in comparison with other algorithms based on multipopulation methods on the moving peaks benchmark.
Description
Keywords
Clustering, Differential evolution, Dynamic optimization problem, Genetic algorithm, Multiple population methods, Particle swarm optimization undetectable dynamism
Citation
Li, C. and Yang, S. (2012) A general framework of multipopulation methods with clustering in undetectable dynamic environments. IEEE Transactions on Evolutionary Computation, 16(4), August 2012, pp. 556-577.
Research Institute
Institute of Artificial Intelligence (IAI)