Recursive nonlinear estimation of nonlinear/non-Gaussian dynamic models
The paper presents a new approach to systematic design of approximate recursive Bayesian estimation of nonlinear and/or non-Gaussian dynamic models. It is based on the use of information measures and Pythagorean geometry of probability spaces. Powerful concepts are introduced: the concept of inaccuracy of the empirical distribution of observed data relative to their theoretical distribution; a Pythagorean-like decomposition of the inaccuracy into a sum of two terms---one conditionally independent of the observed data given the value of a certain statistic, the other independent of the parameter of an approximating exponential family. The Pythagorean decomposition is used to justify a conceptual scheme of approximate estimation. It is shown how one can choose a suitable data statistic that preserves as much information from data as possible. A first-order autoregressive process with Cauchy noise is used to illustrate the method. Recommendations for efficient implementation of the approximate estimation algorithm are given. For the entire collection see [Zbl 0892.00024].
- scientific article; zbMATH DE number 992649
- A nonparametric method of estimating nonlinear dynamical system models
- Nonparametric estimation of nonlinear dynamic systems using semirecursive regression estimates
- Nonparametric recursive estimation in nonlinear ARX-models
- Recursive prediction error parameter estimator for non-linear models
- Recursive Estimation for Some Nonstationary Processes
- Robust recursive estimation in nonlinear time series
- Recursive nonlinear estimation. A geometric approach
- Direct approach to estimation and nonlinear estimation of measurements
- Recursive prediction error identification and scaling of non-linear state space models using a restricted Black box parameterization
- A bayes-closed approximation of recursive non-linear estimation
- scientific article; zbMATH DE number 3930874 (Why is no real title available?)
- scientific article; zbMATH DE number 472982 (Why is no real title available?)
- Recursive Bayesian estimation of autoregressive model with uniform noise using approximation by parallelotopes
- scientific article; zbMATH DE number 4117658 (Why is no real title available?)
- scientific article; zbMATH DE number 1437502 (Why is no real title available?)
- On using Hellinger distance in checking the validity of dynamic approximations
- Recursive Update Filtering for Nonlinear Estimation
- scientific article; zbMATH DE number 4188995 (Why is no real title available?)
- Approximate Bayesian approach to non-Gaussian estimation in a linear model with dependent state and noise vectors
- Recursive nonlinear estimation: A geometric approach
This page was built for publication: Recursive nonlinear estimation of nonlinear/non-Gaussian dynamic models
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4400819)
