A Bayesian Motivated Laplace Inversion for Multivariate Probability Distributions