Abstract:Gaussian approximations are routinely employed in Bayesian statistics to ease inference when the target posterior is intractable. Although these approximations are asymptotically justified by Bernstein-von Mises type results, in practice the expected Gaussian behavior may poorly represent the shape of the posterior, thus affecting approximation accuracy. Motivated by these considerations, we derive an improved class of closed-form approximations of posterior distributions which arise from a new treatment of a third-order version of the Laplace method yielding approximations in a tractable family of skew-symmetric distributions. Under general assumptions which account for misspecified models and non-i.i.d. settings, this family of approximations is shown to have a total variation distance from the target posterior whose rate of convergence improves by at least one order of magnitude the one established by the classical Bernstein-von Mises theorem. Specializing this result to the case of regular parametric models shows that the same improvement in approximation accuracy can be also derived for polynomially bounded posterior functionals. Unlike other higher-order approximations, our results prove that it is possible to derive closed-form and valid densities which are expected to provide, in practice, a more accurate, yet similarly-tractable, alternative to Gaussian approximations of the target posterior, while inheriting its limiting frequentist properties. We strengthen such arguments by developing a practical skew-modal approximation for both joint and marginal posteriors that achieves the same theoretical guarantees of its theoretical counterpart by replacing the unknown model parameters with the corresponding MAP estimate. Empirical studies confirm that our theoretical results closely match the remarkable performance observed in practice, even in finite, possibly small, sample regimes.

A Modified Satterthwaite (1941,1946) Effective Degrees of Freedom Approximation

LIL and the Approximation of Rectangular Sums of B-valued Random Variables when Extreme Terms Are Excluded

Modified score functions for von Mises regressions

A note on the accuracy of adaptive Gauss–Hermite quadrature

Exact Multivariate Tests - A New Effective Principle of Controlled Model Choice

Skewed Bernstein-von Mises theorem and skew-modal approximations

Adaptive scaled unscented transformation for highly efficient structural reliability analysis by maximum entropy method

A further improvement of the quantitative Subspace Theorem

Degrees of Freedom: Search Cost and Self-Consistency

Modified method of moments for generalized Laplace distributions

Correction to "Wasserstein distance estimates for the distributions of numerical approximations to ergodic stochastic differential equations"

Approximate Power for Repeated-Measures ANOVA Lacking Sphericity

A comment on the article "The Schwarz alternating method in solid mechanics" by Alejandro Mota, Irina Tezaur and Coleman Alleman [Comput. Methods Appl. Mech. Engrg. 319 (2017) 1951]

Degrees of Freedom and Model Search

Improved approximations to the Wagner function using sparse identification of nonlinear dynamics

Error of Approximation of Functions Belonging to -Class Through Double Matrix Means

Reduced basis approximation and a posteriori error estimation: applications to elasticity problems in several parametric settings

Modified approximation method for structural failure probability analysis of high-dimensional systems

An efficient approach for high-dimensional structural reliability analysis

Saddlepoint approximation for Student's t-statistic with no moment conditions

An Approximate Quadratic Analysis of Fixture Locating Schemes