计算机科学 ›› 2011, Vol. 38 ›› Issue (7): 231-234.
• 人工智能 • 上一篇 下一篇
张宇山,郝志峰,黄翰
出版日期:
发布日期:
基金资助:
ZHANG Yu-shan,HAO Zhi-feng,HUANG Han
Online:
Published:
摘要: 进化算法的理论研究,如收敛性、时间复杂性研究,是当前的一大热点和难点,有关的理论结果并不多。针对二元进化策略(1+1) ES建立时齐马尔科夫过程模型,利用连续状态马氏过程理论证明了与(1+1) ES相关联的马氏过程在一类连续优化问题中具有指数遍历性,在此基础上证明了(1+1)ES在求解此类优化问题时能以概率1最终找到最优解。所提出的分析方法为进化算法的理论研究提供了一条新思路。
关键词: 进化计算,进化策略,收敛性,连续优化,马尔科夫过程
Abstract: The theoretical investigation to Evolutionary Algorithms, e. g. convergence analysis, runtime analysis, is currently a hot topic, and the related theoretical results arc few despite many experimental results. I}his paper established a homogeneous Markov process model associated with(1+1)ES .It is proved by means of the theory of Markov process with continuous state that the Markov process associated with(1+1) ES has exponential ergodicity in a class of continuous optimization problem,hence the(1+1)ES can finally converge to the optimal solution with probability 1 when solving such a problem. The proposed analytic approach provides a new thought to the theoretical research of evolutionary algorithms.
Key words: Evolutionary computation, Evolution strategy, Convergence, Continuous optimization, Markov process
张宇山,郝志峰,黄翰. 二元进化策略的收敛性分析[J]. 计算机科学, 2011, 38(7): 231-234. https://doi.org/
ZHANG Yu-shan,HAO Zhi-feng,HUANG Han. Convergence Analysis of Two-membered Evolution Strategy[J]. Computer Science, 2011, 38(7): 231-234. https://doi.org/
0 / / 推荐
导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks
链接本文: https://www.jsjkx.com/CN/
https://www.jsjkx.com/CN/Y2011/V38/I7/231
Cited