Small-noise approximation for Bayesian optimal experimental design with nuisance uncertainty

Calculating the expected information gain in optimal Bayesian experimental design typically relies on nested Monte Carlo sampling. When model also contains nuisance parameters, which are parameters that contribute to overall uncertainty of system but no interest framework, this introduces a second inner loop. We propose and derive small-noise approximation for additional The computational cost our method can be further reduced by applying Laplace remaining Thus, we present two methods, double-loop methods. Moreover, demonstrate total complexity these approaches remains comparable case without uncertainty. To assess efficiency three examples, last example includes partial differential equation electrical impedance tomography experiment composite laminate materials.