Sequential Bayesian inference for static parameters in dynamic state\n space models

2014/08/20 by Arnab Bhattacharya, Bhattacharya, Arnab, Simon Wilson +1
Computer Science · Decision Sciences · Engineering · #62L10 #Bayesian Modeling and Causal Inference #Computation (stat.CO) #FOS: Computer and information sciences #Fault Detection and Control Systems #Forecasting Techniques and Applications #G.3

paper · pdf · doi:10.48550/arxiv.1408.4559

openalex publication_date 2014/08/20 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28

Abstract

A method for sequential Bayesian inference of the static parameters of a\ndynamic state space model is proposed. The method is based on the observation\nthat many dynamic state space models have a relatively small number of static\nparameters (or hyper-parameters), so that in principle the posterior can be\ncomputed and stored on a discrete grid of practical size which can be tracked\ndynamically. Further to this, this approach is able to use any existing\nmethodology which computes the filtering and prediction distributions of the\nstate process. Kalman filter and its extensions to non-linear/non-Gaussian\nsituations have been used in this paper. This is illustrated using several\napplications: linear Gaussian model, Binomial model, stochastic volatility\nmodel and the extremely non-linear univariate non-stationary growth model.\nPerformance has been compared to both existing on-line method and off-line\nmethods.\n

Related

• Bayesian Inference in Nonparametric Dynamic State-Space Models
• Nested smoothing algorithms for inference and tracking of heterogeneous multi-scale state-space systems
• Nested Gaussian filters for recursive Bayesian inference and nonlinear\n tracking in state space models
• Approximate Bayesian Computation in State Space Models
• Nested smoothing algorithms for inference and tracking of heterogeneous multi-scale state-space systems