文献类型：期刊文献

英文题名：Estimation of distribution algorithm sampling under gaussian and cauchy distribution in continuous domain

作者：Luo, Na[1,2]; Qian, Feng[1,2]

机构：[1] Automatic Institute, Key Laboratory of Advanced Control and Optimization for Chemical Processes, Ministry of Education, Shanghai, 200237, China; [2] State-Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai, 200237, China

年份：2010

起止页码：1716

外文期刊名：2010 8th IEEE International Conference on Control and Automation, ICCA 2010

收录：EI(收录号：20104213298428)

语种：英文

外文关键词：Evolutionary algorithms - Optimization - Probability density function - Heuristic algorithms

摘要：Estimation of Distribution Algorithm is a new population based evolutionary optimization method and it generates new population from probability distribution model. Like most evolutionary algorithms, it is easy to trap into local optimums. In order to avoid this shortcoming, Gaussian and Cauchy probability density function are mixed as probability distribution model. For continuous problems, a new estimation of distribution algorithm sampling under the mixed model is presented. New individuals are generated not only from Gaussian distribution but sometimes from Cauchy distribution in order to keep diversity. The selection strategy of Gaussian and Cauchy distribution are also discussed. The new algorithm is tested on five benchmark functions and results are compared with basic and estimation of distribution algorithm with Cauchy mutation. ? 2010 IEEE.