We develop a numerical procedure that facilitates efficient likelihood evaluation in applications involving non-linear and non-Gaussian state-space models. The procedure approximates necessary integrals using continuous approximations of target densities. Construction is achieved via efficient...