A1295
Title: Parameter inference for partially observed, implicitly defined simulation models
Authors: Joonha Park - University of Kansas (United States) [presenting]
Abstract: In many applications, a stochastic system is studied using a model implicitly defined via a simulator. A simulation-based parameter inference method is developed for such implicitly defined models where partial or noisy observations are available. The method differs from traditional likelihood-based inference in that it uses a simulation metamodel for the distribution of a log-likelihood estimator, which is built on a local asymptotic normality (LAN) property. The use of a simulation metamodel enables scalable parameter estimation and uncertainty quantification with increasing data size. The method is demonstrated using numerical examples, including a mechanistic model for the population dynamics of an infectious disease.
