1: \begin{abstract}

2: We develop a model-based empirical Bayes approach to variable selection problems

3: in which the number of predictors is very large, possibly much larger than the number

4: of responses (the so-called “large p, small n” problem). We consider the multiple linear 

5: regression setting, where the

6: response is assumed to be a continuous variable and it is a linear function of

7: the predictors plus error. The explanatory variables in the linear model can have a positive

8: effect on the response, a negative effect, or no effect. We model the effects

9: of the linear predictors as a three-component mixture

10: in which a key assumption is

11: that only a small (unknown) fraction of the candidate predictors have a non-zero effect on

12: the response variable. By treating the coefficients as random effects we develop

13: an approach that is computationally

14: efficient because the number of parameters that have to be estimated is small, and

15: remains constant regardless of the number of explanatory variables. 

16: The model parameters are estimated using the EM algorithm which

17: is scalable and leads to significantly faster convergence, compared with simulation-based methods.

18: \end{abstract}

19: