Estimating percentiles of black-box deterministic functions with randominputs is a challenging task when the number of function evaluations isseverely restricted, which is typical for computer experiments. This articleproposes two new sequential Bayesian methods for percentile estimation based onthe Gaussian Process metamodel. Both rely on the Stepwise Uncertainty Reductionparadigm, hence aim at providing a sequence of function evaluations thatreduces an uncertainty measure associated with the percentile estimator. Theproposed strategies are tested on several numerical examples, showing thataccurate estimators can be obtained using only a small number of functionsevaluations.
展开▼
