基于智能优化算法的模糊软子空间聚类方法

为解决选定特征上的聚类问题和模糊C-均值聚类存在的初始值敏感、易陷入局部最优的问题,提出了一种基于改进萤火虫算法的模糊软子空间聚类方法。该方法在模糊C-均值聚类算法的基础上,采用基于数据可靠性的k-均值算法中特征权值的计算方法,并结合萤火虫算法的全局搜索能力对所有的特征子空间进行搜索；设计了一种目标函数来对聚类结果和子空间所包含的特征维进行评估,并利用目标函数改进了萤火虫算法的搜索公式。实验结果表明,该方法能有效地收敛于全局最优解,具有良好的聚类效果和抗噪性。

Fuzzy Soft Subspace Clustering Method Based on Intelligent Optimization Algorithm

To solve the issue of clustering on selected characteristics and the problems that fuzzy C-means is sensitive to initial value and easy to fall into local optimum,a new fuzzy subspace clustering method based on improved firefly algorithm was proposed.Based on fuzzy C-means clustering algorithm,the method uses the way to calculate feature weighting in reliability-based k-means algorithm,and combines with the global search capability of firefly algorithm to search for all the subspace.An objective function was designed to evaluate the clustering results and feature-dimension included in subspace,and it was adopted to improve the searching formula of firefly algorithm.Experimental results show that the proposed clustering method can effectively converge to the global optimal solution,and has good clustering effect and noise immunity.