摘要
针对粗糙模糊聚类算法对初值敏感、易陷入局部最优和聚类性能依赖阈值选择等问题,提出一种混合蛙跳与阴影集优化的粗糙模糊聚类算法(SFLA-SRFCM).通过设置自适应调节因子,以增加混合蛙跳算法的局部搜索能力;利用类簇上、下近似集的模糊类内紧密度和模糊类间分离度构造新的适应度函数;采用阴影集自适应获取类簇阈值.实验结果表明,SFLA-SRFCM算法是有效的,并且具有更好的聚类精度和有效性指标.
For the problem that the rough fuzzy clustering algorithm is sensitive to the initial value, easy to fall into a local optimal solution, and the clustering performance of algorithm depends on the selection of threshold, a rough fuzzy clustering algorithm based on the shuffled frog leaping algorithm and shadowed sets（SFLA-SRFCM） is proposed. The adaptive factor is developed to enhance the local search ability, the within cluster tighness and the between cluster scatter of fuzzy lower approximate sets and fuzzy upper approximate sets are used to construct a new fitness function. Shadowed sets are applied to obtain the threshold adaptively. Experimental results show that SFLA-SRFCM is effective and has better clustering accuracy and validity index.
出处
《控制与决策》
EI
CSCD
北大核心
2015年第10期1766-1772,共7页
Control and Decision
基金
国家自然科学基金项目(61363027)
广西自然科学基金项目(2012GXNSFAA053225)
关键词
粗糙集
阴影集
粗糙模糊聚类
混合蛙跳算法
rough sets
shadowed sets
rough fuzzy clustering
shuffled frog leaping algorithm
