Relational fuzzy c-lines derived from kernel fuzzy c-lines

In this paper, three linear fuzzy clustering algorithms are proposed for relational data based on kernel fuzzy c-means, in which the prototypes of clusters are given by lines spanned in a feature space denned by the kernel which derived from a given relational data. The proposed algorithms contrast the conventional method in which the prototypes of clusters are given by lines spanned by two representative objects. Through numerical examples, it is shown that the proposed algorithms can capture local sub-structures in relational data.