基于遗传进化的聚类分析新方法  

A New Chemical Clustering Analysis Method Based on Genetic Evolution Algorithm

作　　者：陈文灿[1] 崔卉[1] 张林 梁逸曾 

机构地区：[1]湖南大学化学化工学院化学计量学与化学传感技术研究所,长沙410082

出　　处：《高等学校化学学报》1997年第2期196-201,共6页Chemical Journal of Chinese Universities

基　　金：国家自然科学基金;国家教育委员会霍英东基金

摘　　要：提出了一种基于遗传进化策略的聚类分析新方法，将聚类问题转换为实数空间的优化问题．用遗传策略在连续的实数空间寻找最优类中心，既可避免求解组合优化问题，减少计算量，又可避免在优化过程中陷入局部最优．此遗传聚类方法用于对4组实际数据进行硬分类和模糊分类，结果令人满意．Genetic evolution strategy (GEs) was used in chemical clustering analysis. The clustering objective was formulated as a function of real-valued parameters. Combinatorial optimization usually encountered in clustering analysis, which is commonly time-consuming in computation, can be avoided, and therefore, the amount of computation can be significantly reduced. Global optimum might be achieved with the help of genetic algorithm. Results obtalned from four real chemical data sets substantiate the practicability of the method proposed in the paper.

关 键 词：聚类分析 模糊分类 遗传进化策略 模式识别 

分 类 号：O235[理学—运筹学与控制论] O212.4[理学—数学]