改进的遗传模糊聚类算法  

An Improved Hybrid Cluster Algorithm

作　　者：董云影[1] 张运杰[2] 畅春玲[3] 

机构地区：[1]大连水产学院理学院,辽宁大连116026 [2]大连海事大学数学系,辽宁大连116026 [3]沈阳建筑大学理学院,辽宁沈阳110000

出　　处：《模糊系统与数学》2005年第2期128-133,共6页Fuzzy Systems and Mathematics

基　　金：大连海事大学科研基金资助项目

摘　　要：对基于遗传算法的FCM(模糊c-均值法)聚类算法进行了改进,能更好地把遗传算法的全局搜索能力和FCM的局部搜索能力结合起来。实验结果表明,这种改进的算法在分类正确率和稳定性上优于[1]和[3]中的方法;收敛速度和对初值的敏感性都明显优于FCM。<Abstrcat> In this paper, we propose an improved hybrid cluster algorithm combined genetic (algorithms) with FCM(fuzzy c-means algorithms), which can achieve the combination of genetic (algorithm's) global search ability and FCM's local search ability. Experiments show that this modified algorithm's accuracy and stability are better than the algorithms in [1] and [3]. It is also observed (that) the converging speed and the sensibility to initial value in our method are definitely better than FCM.

关 键 词：遗传算法 FCM 模糊聚类 

分 类 号：O159[理学—数学]