1: \begin{abstract}

2: %		%	In this paper, we study the problem of personalized attribute learning, which aims to predict individual annotations for visual attributes. We design a hierarchal    

3: %		Traditionally, most of the existing attribute learning methods are trained based on the consensus of annotations aggregated from a limited number of annotators. However, it is hard for different people to come to an agreement, even when evaluating the same attribute for the same object, due to the significant diversity of their evaluation criteria and their comprehension of the attribute word. In this paper, we propose a novel multi-task method to incorporate the personalized diversity into attribute learning. To capture the evolution from a highly popular opinion of the mass to highly personalized choices that are special for each person, we propose a multi-level decomposition in our model where a globally shared factor, a user-feature co-grouping factor, and a personalized factor are simultaneously learned. Meanwhile, for personalized learning methods, ranking prediction is much more important than accurate classification. This motivates us to employ an Area Under ROC Curve (AUC) based loss function to improve our model. On top of the AUC-based loss, we propose an efficient method to evaluate the loss and gradients. Theoretically, we propose a novel closed-form solution for one of our non-convex subproblem, which leads to provable convergence behaviors. Furthermore, we also provide a generalization bound to guarantee a reasonable performance.  Finally, empirical analysis consistently speaks to the efficacy of our proposed method.

4: %	\end{abstract}

5: