Fuzzy Fisher Criterion Based Semi-Fuzzy Clustering Algorithm
Cao Su-qun①② Wang Shi-tong① Chen Xiao-feng① Xie Zhen-ping① Deng Zhao-Hong①
①(School of Information, Jiangnan University, Wuxi 214122, China) ②(Department of Mechanical Engineering, Huaiyin Institute of Technology, Huaian 223001, China)
Abstract:The robust Fuzzy Fisher Criterion based Semi-Fuzzy Clustering Algorithm (FFC-SFCA) for linearly separable data is presented in this paper. FFC-SFCA incorporates Fisher discrimination method with fuzzy theory using fuzzy scatter matrix. By iteratively optimizing the fuzzy Fisher criterion function, the final clustering results are obtained. FFC-SFCA exhibits its robustness and capability to obtain well separable clustering results. In addition, optimal discriminant vector and threshold of classifier can also be figured out. The experimental results for artificial and real datasets demonstrate its validity and distinctive superiority over the two conventional clustering algorithms.