This paper provides an alternative approach to penalized regression for model selection in the context of high-dimensional linear regressions where the number of covariates is large, often much larger ...