1: \begin{abstract}

2: Multi-task learning is a paradigm, where multiple tasks are jointly learnt.

3: Previous multi-task learning models usually treat all tasks and instances per task equally during learning.

4: Inspired by the fact that humans often learn from easy concepts to hard ones in the cognitive process, in this paper, we propose a novel multi-task learning framework that attempts to learn the tasks by simultaneously taking into consideration the complexities of both tasks and instances per task.

5: We propose a novel formulation by presenting a new task-oriented regularizer that can jointly prioritize tasks and instances.

6: Thus it can be interpreted as a self-paced learner for multi-task learning.

7: An efficient block coordinate descent algorithm is developed to solve the proposed objective function, and the convergence of the algorithm can be guaranteed.

8: Experimental results on the toy and real-world datasets demonstrate the effectiveness of the proposed approach, compared to the state-of-the-arts.

9: \end{abstract}

10: