Probabilistic multi-label classification with sparse feature learning

摘要: Multi-label classification is a critical problem in many areas of data analysis such as image labeling and text categorization. In this paper we propose probabilistic multi-label model based on novel sparse feature learning. By employing an individual sparsity inducing l1-norm group l2,1-norm, the proposed has capacity capturing both label interdependencies common predictive structures. We formulate norm regularized learning non-smooth convex optimization problem, develop fast proximal gradient algorithm to solve it for optimal solution. Our empirical study demonstrates efficacy method set tasks given limited number labeled training instances.