遗传与C-均值混合算法用于聚类分析  

CLUSTERING BY HYBRID GENETIC/C-MEANS ALGORITHM

作　　者：王涛[1] 沈谦[2] 朱明星 张良震[2] 

机构地区：[1]中国科学技术大学电子科学与技术系,合肥230026 [2]安徽大学电子工程与信息科学系,合肥230039

出　　处：《模式识别与人工智能》1999年第1期98-103,共6页Pattern Recognition and Artificial Intelligence

基　　金：安徽省教委自然科学基金

摘　　要：本文提出了一个将遗传算法与C-均值算法相结合的混合算法,并根据需要修改了交叉算子和变异算子。实验结果表明,该算法较好地避免了局部极小,同时收敛速度较快。C-means algorithm which is one of the most, popular clustering technique, usually obtains local minimum results. That shortcoming can be settled by the proposed algorithm with the help of genetic algorithm which is powerful in finding the global or near global optimal solutions of optimization problems. The crossover operator and mutation operator are modified to the requirements. Experimental results show that the proposed method with less iterations can efficiently avoid falling into local minima.

关 键 词：聚类分析 C-均值算法 遗传算法 

分 类 号：O212.4[理学—概率论与数理统计]