An efficient procedure for feature selection for generalized linear models with L0 penalty, including linear, logistic, Poisson, gamma, inverse Gaussian regression. Adaptive ridge algorithms are used to fit the models.
