基于遗传算法与FCSS相结合的模糊球壳聚类算法  

Fuzzy C-spherical shell cluster algorithm based on genetic technique and FCSS

作　　者：惠周利[1] 杨明[1] 潘晋孝[1] 

机构地区：[1]中北大学数学系,山西太原030051

出　　处：《传感器与微系统》2008年第12期109-111,共3页Transducer and Microsystem Technologies

基　　金：山西省自然科学基金资金项目(20051042);山西省研究生教育创新资金资助项目(20061024)

摘　　要：模糊球壳聚类(FCSS)算法广泛地应用于模式识别与机器学习等领域。由于其采用基于梯度法和交替寻优策略,对初始化比较敏感,容易陷入局部极值点,从而影响聚类效果。将现代全局优化方法之一的遗传算法(GA)与FCSS算法相结合,得到一种新的球壳聚类算法GA-FCSS。数值实验表明:新方法对球壳形数据有令人满意的聚类效果。Fuzzy C-spherical shell cluster algorithm（FCSS） is widely applied to pattern recognition and machine learning. The gradient method and alternative optimization strategy are adopted, its shortcoming is the sensibility to initial value and it is easy to run into a local optimum, so affecting the clustering effect. A new algorithm called GA-FCSS,which integrated with the FCSS algorithm and the genetic algorithm（GA） is put forward. Simulated and comparative experiments show the feasibility and availability of the proposed method.

关 键 词：模糊聚类 模糊球壳聚类算法 遗传算法 

分 类 号：O235[理学—运筹学与控制论]