【摘要】 针对现有直觉模糊集聚类方法存在计算量大、数据失真和易陷于局部最优等问题,提出基于新直觉模糊相似度量的直觉模糊谱聚类算法。首先定义了新的直觉模糊相似度量方法,然后基于该方法构造了直觉模糊相似度矩阵,根据直觉模糊相似度矩阵求解非规范Laplacian矩阵,在此基础上构建特征矩阵,再使用k-means算法对特征矩阵进行聚类。最后在数值算例上的应用证明了所提出算法的可行性和有效性。更多还原

【Abstract】 The existing clustering algorithms of inituitionistic fuzzy sets need a large amount of computation efforts, or brings the data distortion, or even may fall into the local optimal solutions. A novel spectral algorithm based on inituitionistic fuzzy sets is proposed. Firstly, a novel inituitionistic fuzzy similarity measure is defined. Next use it to construct an intuitionistic fuzzy similarity measure matrix, by which we present a spectral clustering algorithm to cluster intuitionistic fuzzy information. At last, a numerical examples shows the feasibility and validity of this algorithm.更多还原