Fast Bayesian experimental design: Laplace-based importance sampling for the expected information gain

TL;DR

This paper introduces a Laplace-based importance sampling method to efficiently compute expected information gain in Bayesian experimental design, reducing computational cost and overcoming numerical issues of traditional double-loop Monte Carlo methods.

Contribution

The paper develops a Laplace-based importance sampling approach for Bayesian experimental design, providing optimal parameter choices and demonstrating improved efficiency over existing methods.

Findings

Significant reduction in computational cost compared to classical methods

Effective in both linear and nonlinear scalar problems

Applicable to complex sensor placement in electrical impedance tomography

Abstract

In calculating expected information gain in optimal Bayesian experimental design, the computation of the inner loop in the classical double-loop Monte Carlo requires a large number of samples and suffers from underflow if the number of samples is small. These drawbacks can be avoided by using an importance sampling approach. We present a computationally efficient method for optimal Bayesian experimental design that introduces importance sampling based on the Laplace method to the inner loop. We derive the optimal values for the method parameters in which the average computational cost is minimized according to the desired error tolerance. We use three numerical examples to demonstrate the computational efficiency of our method compared with the classical double-loop Monte Carlo, and a more recent single-loop Monte Carlo method that uses the Laplace method as an approximation of the return value of the inner loop. The first example is a scalar problem that is linear in the uncertain parameter. The second example is a nonlinear scalar problem. The third example deals with the optimal sensor placement for an electrical impedance tomography experiment to recover the fiber orientation in laminate composites.