Tommie Catanach

We propose a sequential Monte Carlo estimator for expected information gain that reduces computation in Bayesian optimal experiment design using backward tempering.