1: \begin{abstract}

2: %Symmetric non-negative matrix factorization (SymNMF) is a  low rank factorization method for graph clustering, which decomposes the input Laplacian graph weight matrix into the product of a non-negative matrix and its transpose.

3: %Different from the traditional  graph clustering methods that usually need a post-processing step like K-means to produce the finial clustering result, the  factorized matrix in SymNMF can directly indicates the clustering membership. 

4: %%Due to the non-negativity of resulting matrix, it can act as a clustering membership matrix 

5: %%and can  produce the clustering membership directly without post-processing method like K-means. 

6: %However, it still suffers from two main limitations, i.e., $i)$, once the initial graph is determined, the clustering process is no longer related to the input data, however, the initial similarity graph maybe not always the best choice; and $ii)$, SymNMF is a feed-forward process,  where the clustering results generated by SymNMF that can provide the discriminative information can not affect the SymNMF process. 

7: %To solve the above limitations, in this paper, we propose an adaptive graph learning guided SymNMF model (AGLSymNMF), where the graph weight matrix is adaptively updated by compromising the information from the input data, the initial Laplacian graph and the clustering result. 

8: %Moreover, to further make use of the discriminative information from clustering result, we add an orthogonal constraint on the resulting matrix.

9: %% extend the proposed AGLSymNMF to enhance the feedback from 

10: %%Moreover, we extend the proposed AGLSymNMF to a kernel version to exploit the  non-liner information in the input data. 

11: %The proposed model is efficiently solved with convergence theoretically guaranteed. We apply the proposed model to 10 benchmark datasets and  the advantages of the proposed model is demonstrated by comparing with state-of-the-art clustering methods under various metrics. 

12: %\end{abstract}

13: