1: \begin{abstract}

2:   Multi-label learning is usually used to mine the correlation 

3: between features and labels, 

4: and feature selection can retain as much information as possible 

5: through a small number of features.

6: $\ell_{2,1}$ regularization method can get sparse coefficient matrix, 

7: but it can not solve multicollinearity problem effectively. 

8: The model proposed in this paper can obtain the most relevant few features 

9: by solving the joint constrained optimization problems of $\ell_{2,1}$ 

10: and $\ell_{F}$ regularization.

11: In manifold regularization, we implement random walk strategy based on 

12: joint information matrix, and get a highly robust neighborhood graph.

13: In addition, we given the algorithm for solving the model and proved 

14: its convergence.

15: Comparative experiments on real-world data sets show that 

16: the proposed method outperforms other methods.

17: \end{abstract}

18: