Particle Methods for On-Line Parameter Estimation in Non-Linear Non-Gaussian State-space Models

OMI Seminar Series

Nonlinear non-Gaussian state-space models are ubiquitous in statistics, econometrics and related fields. Particle methods, also known as Sequential Monte Carlo (SMC) methods, provide reliable numerical approximations to the associated state inference problems. However, in most applications, the state-space model of interest also depends on unknown static parameters that one would like to estimate on-line. We present new particle algorithms to perform on-line maximum likelihood static parameter estimation. These algorithms are provably numerically stable and do not suffer from the particle path degeneracy problem.

Joint work with Pierre Del Moral (INRIA Bordeaux) and Sumeet Singh (Cambridge University)


Arnaud Doucet (Oxford-Man Institute)

Tuesday, November 1, 2011 - 14:15
to 15:15