Applications of filtering and data assimilation arise in
engineering, geosciences, weather forecasting, and many other areas where one has to make predictions based on uncertain models supplemented by a stream of data with noise. For nonlinear problems filtering can be very expensive. In this talk, a particle-based nonlinear filtering scheme will be presented. This algorithm is based on implicit sampling, a new sampling technique related to chainless Monte Carlo method. Its main features are that the posterior densities are represented by pseudo-Gaussians and a resampling based on normalization constants. This filter is designed to focus particle paths sharply so as to reduce the number of particles needed for nonlinear problems. Examples will be given.
host: Kui Ren