摘要
针对模糊C-均值聚类算法对初始化分类参数的选择比较敏感而导致分类结果差异性较大的不足,提出基于模糊C-均值聚类目标函数相对权重系数的偏导函数进行数据分类效果好坏的评价。实验结果表明,该文定义的分类效果评价方法是可行的。
Based on the shortage of fuzzy c-means algorithm which initialized classification parameter is sensitivity to data classifying quality,and different initialized classification parameters generate classifying result with bigger otherness.A new evaluating criterion based on partial derivative of objective with respect to its weight coefficient is put forward to assess data classifying quality in this paper.Experimental results show that an evaluating criterion proposed in this paper is feasible.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第25期171-172,222,共3页
Computer Engineering and Applications
基金
国家自然科学基金项目:聚类有效性问题的研究(编号:69972041)资助
关键词
模糊C-均值聚类
目标函数
分类效果
fuzzy c-means algorithm,objective function,classifying quality
