The Linearized Alternating Direction Method Of Multipliers For Mixed Matrix Regression Model

In modern scientific applications,data sets are often in matrix form.In which,there is a coefficient matrix needs to be estimated.Zhou and Li proposed the matrix re-gression models,but they mainly focused on the matrix without vector variables.In this paper,we consider the mixed matrix regression model.It can deal with low rank matrix data and sparse vector data meanwhile.To make the mixed matrix regression model practically feasible,we propose a linearized alternating direction method of multipliers(LADMM)and establish its global convergence.Some numerical studies are reported to demonstrate the efficiency of our proposed method.