Abstract
In this study, we are concerned with a concept of consensus-driven fuzzy clustering whose objective is to reconcile a structure developed for patterns in some data set with the structural findings already available for other related data sets (where these data sets are reflective of the same phenomenon which has led to the generation of the original patterns). The results of fuzzy clustering are provided in the form of prototypes and fuzzy partition matrices. Given this form of representation of granular results (clusters), we develop a suitable communication scheme using which consensus could be established in an effective manner. Here, we consider proximity matrices induced by the corresponding partition matrices. An overall optimization scheme is presented in detail along with a way of forming a pertinent criterion governing an intensity of collaboration between the data driven- and knowledge oriented hints guiding the process of consensus formation. Several illustrative numeric examples, using both synthetic data and the data coming from publicly available machine learning repositories are also included.