Fast semi-supervised evidential clustering

Semi-supervised clustering is a constrained technique that organizes collection of unlabeled data into homogeneous subgroups with the help domain knowledge expressed as constraints. These methods are, most time, variants popular k-means algorithm. As such, they are based on criterion to minimize. Amongst existing semi-supervised clusterings, Evidential Clustering (SECM) deals problem uncertain/imprecise labels and creates credal partition. In this work, new heuristic algorithm, called SECM-h, presented. The proposed algorithm relaxes constraints SECM in such way optimization solved using Lagrangian method. Experimental results show largely improves execution time while accuracy maintained.