In this paper,we focus on the problem of sparse reconstruction.As is proposed,greedy inverse scale space(GISS)is a greedy approach for sparse signal reconstruction in compressive sensing and it is proved to be efficient and accurate compared with other greedy and relaxed methods.We extend the GISS method for solving the sparse precision matrix reconstruction problem.Numerical tests on simulation examples show the advantages of the GISS method.Finally,we apply the GISS method to the diagnosis of Alzheimer's Disease(AD),comparing the classification effect between covariance matrix and precision matrix. |