A relational version variant of the evidential c-means (ECM) algorithm. As ECM, it is based on the concept of credal partition, extending those of hard, fuzzy and possibilistic ones. To derive such a structure, a suitable objective function is minimized using an alternating optimization algorithm similar to the fuzzy or possibilistic c-means algorithms. In contrast to ECM, RECM takes as input a proximity matrix. The method is described in:

M.-H. Masson and T. Denoeux. RECM: Relational Evidential c-means algorithm. Pattern Recognition Letters, Accepted for publication, 2009.. pdf

Download