Letters: Infinite-dimensional sampling inference algorithm for linear dynamic system

Authors:
Juyang Lei;Ke Huang;Haixiang Xu;Xizhi Shi
Affiliations:
State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University, Shanghai 200240, China;State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University, Shanghai 200240, China;State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University, Shanghai 200240, China;State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University, Shanghai 200240, China
Venue:
Neurocomputing
Year:
2009

Citing 3
Cited 0

Dynamic bayesian networks: representation, inference and learning

Dynamic bayesian networks: representation, inference and learning
Graphical models for visual object recognition and tracking

Graphical models for visual object recognition and tracking
Iterative algorithms for state estimation of jump Markov linearsystems

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.01

Visualization

Abstract

The Kalman filter (KF) models the propagation of uncertainty for a dynamic system where the noise distribution is Gaussian. This letter mainly explores the property of uncertainty propagation in the case where the noise property is unknown. The Dirichlet process mixture (DPM) model is employed to construct a general estimator of the noise distribution. Under the framework of nonparametric Bayes, we use the block sampling and KF techniques to approximate the posterior distribution of the noise. The simulation experiment shows that the proposed algorithm is efficient.