Abstract:Traditional approaches to pattern recognition tasks normally consider only the unilabel classification problem, that is, each observation (both in the training and test sets) has one unique class label associated to it. Yet in many real-worl...
