We propose a multi-fidelity simulation-based inference method based on multilevel Monte Carlo for computationally costly simulators.

A near-optimal quantum-accelerated Monte Carlo estimator for nested expectations