ProPML: Probability Partial Multi-label Learning

Abstract--Partial Multi-label Learning (PML) is a type of weakly supervised learning where each training instance corresponds to a set of candidate labels, among which only some are true. ProPML outperforms existing approaches, especially for high noise in a candidate set. Pineapple Deep neural networks are highly effective in many practical applications. However, their success is heavily dependent on the availability of a large dataset with accurate labeling. Figure 1: In partial multiple-label learning, each training instance Obtaining such datasets is challenging due to the cost and corresponds to a set of candidate labels.