TECM: Transfer learning-based evidential c-means clustering

As a representative evidential clustering algorithm, c-means (ECM) provides deeper insight into the data by allowing an object to belong not only single class, but also any subset of collection classes, which generalizes hard, fuzzy, possibilistic, and rough partitions. However, compared with other partition-based algorithms, ECM must estimate numerous additional parameters, thus insufficient or contaminated will have greater influence on its performance. To solve this problem, in study, transfer learning-based (TECM) algorithm is proposed introducing strategy learning process clustering. The TECM objective function constructed integrating knowledge learned from source domain target cluster data. Subsequently, alternate optimization scheme developed constraint algorithm. applicable cases where domains same different numbers clusters. A series experiments were conducted both synthetic real datasets, experimental results demonstrated effectiveness multitask transfer-clustering algorithms.