Ensemble state estimators in high-dimensions

Optus-Curtin Centre of Excellence in AI Seminar Series

Abstract: Ensemble filtering is a data assimilation method for filtering in high-dimensional state-space models arising in, e.g., fluid mechanics, weather forecasting, and geophysical sciences. In this talk I will introduce ensemble/particle filtering methods. I will relate classical particle Bayesian filters with so-called ensemble state estimators. I will present one particular ensemble filtering method as a mean-field McKean-Vlasov-type particle interpretation of a certain nonlinear diffusion.
Bio: A.N. Bishop currently holds a joint position with the School of Biomedical Engineering at UTS and the CSIRO at the Australian Technology Park in Sydney. Until recently, Dr. Bishop held an ARC DECRA Fellowship (originally at the ANU for 3 years, and then at UTS for 3 years). Prior to joining UTS, Dr. Bishop held academic positions at the Australian National University (ANU) in Canberra, Australia, and the Royal Institute of Technology (KTH) in Stockholm, Sweden.