CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1180
Title: Inference for diffusion processes from observations of passage times Authors:  Moritz Schauer - Leiden University (Netherlands) [presenting]
Abstract: A diffusion process with unknown parameters is indirectly observed with observations being a sequence of passage times. We introduce a method to accurately simulate the conditional diffusion process given the observed random times. A change of measure is applied to a proposal process with a guiding drift term derived from an approximation of the conditional process. The method is used in a Markov chain Monte Carlo procedure to sample from the joint posterior distribution of the unobserved diffusion trajectory and the model parameters given the observed random times. This is illustrated fitting a diffusion model for neuronal spike generation to observations of spike times.