Optimal quantization methods for nonlinear filtering with discrete-time observations
Pagès, Gilles ; Pham, Huyên
Bernoulli, Tome 11 (2005) no. 1, p. 893-932 / Harvested from Project Euclid
We develop an optimal quantization approach for numerically solving nonlinear filtering problems associated with discrete-time or continuous-time state processes and discrete-time observations. Two quantization methods are discussed: a marginal quantization and a Markovian quantization of the signal process. The approximate filters are explicitly solved by a finite-dimensional forward procedure. A posteriori error bounds are stated, and we show that the approximate error terms are minimal at some specific grids that may be computed off-line by a stochastic gradient method based on Monte Carlo simulations. Some numerical experiments are carried out: the convergence of the approximate filter as the accuracy of the quantization increases and its stability when the latent process is mixing are emphasized.
Publié le : 2005-10-14
Classification:  Euler scheme,  Markov chain,  nonlinear filtering,  numerical approximation,  stationary signal,  stochastic gradient descent,  vector quantization
@article{1130077599,
     author = {Pag\`es, Gilles and Pham, Huy\^en},
     title = {Optimal quantization methods for nonlinear filtering with discrete-time observations},
     journal = {Bernoulli},
     volume = {11},
     number = {1},
     year = {2005},
     pages = { 893-932},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1130077599}
}
Pagès, Gilles; Pham, Huyên. Optimal quantization methods for nonlinear filtering with discrete-time observations. Bernoulli, Tome 11 (2005) no. 1, pp.  893-932. http://gdmltest.u-ga.fr/item/1130077599/