糖心原创

Speaker's Name: Andrew Golightly
Speaker's Affiliation: Durham University
Speaker's Research Theme(s): Statistics and Probability
Abstract:
In this work we present a new method for performing Bayesian inference for stochastic kinetic models represented by Markov jump processes. We introduce the Frankenfilter, a principled, partially alive particle filter that targets a user-defined amount of success (such as the number of matches in the case of exactly observed states) whilst fixing lower and upper bounds on the number of simulations. The method produces unbiased estimators of the likelihood, suitable for pseudo-marginal Metropolis-Hastings (PMMH). Using applications arising in computational biology and epidemiology, we demonstrate that PMMH with the Frankenfilter is more robust to outliers and poorly specified initial parameter values than PMMH using standard particle filters and is typically at least 2-3 times more efficient. We also provide advice for choosing the amount of success.

Venue: Monica Partridge B03