1: \begin{abstract}

2: In  recent studies on sparse modeling, $l_q$ ($0<q<1$) regularized

3: least squares regression ($l_q$LS) has received considerable

4: attention due to its superiorities on sparsity-inducing and

5: bias-reduction over the convex counterparts. In this paper, we

6: propose a  \underline{Ga}uss-Seidel \underline{i}terative

7: \underline{t}hresholding \underline{a}lgorithm (called GAITA) for

8: solution to this problem. Different from the classical iterative

9: thresholding algorithms using the Jacobi updating rule, GAITA takes

10: advantage of the Gauss-Seidel rule to update the coordinate

11: coefficients. Under a mild condition, we can justify that the

12: support set and sign of an arbitrary sequence generated by GAITA

13: will converge within finite

14:   iterations. This convergence property together with the

15: Kurdyka-{\L}ojasiewicz property of ($l_q$LS) naturally yields the

16: strong convergence of GAITA  under the same condition as above,

17: which is generally weaker than the  condition for the convergence of

18: the classical iterative thresholding algorithms. Furthermore, we

19: demonstrate that GAITA converges to a local minimizer under certain

20: additional conditions. A set of numerical experiments are provided

21: to show the effectiveness, particularly, much faster convergence of

22: GAITA as compared with the classical iterative thresholding

23: algorithms.

24: \end{abstract}