Evidential prototype-based clustering based on transfer learning

In some real clustering tasks, the data may be sparse and uncertain. Although there is usually useful knowledge in related scenes, among different domains often of great inconsistency. A new unsupervised transfer learning method proposed framework belief functions to handle insufficiency uncertain problems simultaneously. Firstly, under assumption that source target have same number clusters, Transfer Evidential C Means (TECM) developed by incorporating idea evidential clustering. novel objective function designed employ cluster prototypes as references guide process on target. Furthermore, ETECM, an extended version TECM, also introduced for situation two numbers clusters. Some experiments conducted synthetic real-world sets demonstrate advantages TECM ETECM.