面向权重的可能性MKFC算法  

Weight-oriented possibilistic MKFC algorithm

作　　者：唐益明[1] 宋小成 任福继[1] 丰刚永 TANG Yi-ming;SONG Xiao-cheng;REN Fu-ji;FENG Gang-yong(School of Computer and Information,Hefei University of Technology,Hefei 230601,China)

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230601

出　　处：《微电子学与计算机》2020年第12期6-11,共6页Microelectronics & Computer

基　　金：国家自然科学基金(61673156,61432004,U1613217,61672202);安徽省自然科学基金(1408085MKL15,1508085QF129);中国博士后科学基金(2014T70585);国家“八六三”高技术研究发展计划基金(2012AA011103)。

摘　　要：基于多核的模糊聚类(MKFC)是当前聚类领域的最新热点,但是其很难通过人工方式确定在组合中核函数权重大小,并更好地调整所使用的不同内核函数的权重比.为了解决以上问题,设计了自动称量MKFC算法的提案.首先,给出了该算法目标函数的核心公式及其内在思想;其次,给出了Mercer核函数及多核距离的计算方法;然后给出了核函数权重的自动计算模式,由此形成了所提算法的流程;最后,在8个UCI数据库中进行测试,凭借与5种算法的分析对比,发现所提算法是较为理想的聚类方法.Multi-kernel fuzzy clustering(MKFC)is a hot topic in the field of clustering.However,it is difficult to manually determine the weight of kernel function in the combination,and difficult to allocate the weight of kernel function ideally.Focusing on such problem,a weight-oriented possibilistic MKFC algorithm was put forward,which is based on multi-kernel mechanism,possibilistic clustering and automatic weight calculation.Firstly,the core formula of the objective function and its inner idea were given.Secondly,the calculation methods ofthe Mercer kernel function and multi-kernel distance were given.Then the automatic calculation mode of kernel function weight was given and the flow of the proposed algorithm was formed.Finally,the proposed algorithm is tested in 8 UCI databases,and compared with the 5 algorithms.It is found that the proposed algorithm is an ideal clustering method.

关 键 词：聚类 模糊C均值 可能性聚类 核函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]