Collaborative Evidential Clustering

Abstract

Different companies may not be allowed to treat data together given restrictions of security, privacy or other technical reasons. In order to make better use of information from different sources, clustering algorithms based on collaboration mechanisms have been widely used. We propose the concept of collaborative evidential clustering under the framework of evidence theory. The key point is to establish collaboration among the credal partition matrices of each data site to meet the data confidentiality requirements. Considering the problems of excessive information interaction and insufficient information interaction, we design single-step and multi-step collaborative evidential clustering algorithms. Our algorithms were validated on real data sets.