No. 931

Title:

(originally in Japanese)
Parameter Estimattion of State Space Models by Recursive Grid Search for Monte Carlo Filter

Author(s):

Sato, Hideaki (Mizuho Trust and Banking Co., Ltd.)

Key words:

State Space Model; Monte Calro Filter; Grid Search.

Abstract:

    Parameter estimation of state space models by monte calro filter is often difficult. In this Paper, we develop an efficient grid search , RGS(Recursive Grid Search for Monte Carlo Filter) method,that iterates setting a new searching point by using state particles and estimates. Then we apply RGS method to several state space models, and confirm that it can estimate parameters efficiently.