## Mixture Kalman Filters (2000)

Venue: | J. R. Statist. Soc. B |

Citations: | 151 - 5 self |

### BibTeX

@ARTICLE{Chen00mixturekalman,

author = {Rong Chen and Jun S. Liu},

title = {Mixture Kalman Filters},

journal = {J. R. Statist. Soc. B},

year = {2000},

volume = {62},

pages = {493--508}

}

### Years of Citing Articles

### OpenURL

### Abstract

In treating dynamic systems, sequential Monte Carlo methods use discrete samples to represent a complicated probability distribution and use rejection sampling, importance sampling, and weighted resampling to complete the on-line "filtering" task. In this article we propose a special sequential Monte Carlo method, the mixture Kalman filter, which uses random mixture of normal distributions to represent a target distribution. It is designed for on-line estimation and prediction of conditional and partial conditional dynamic linear models, which are themselves a class of widely used nonlinear system and also serve to approximate many other nonlinear systems. Compared with a few available filtering methods including Monte Carlo ones, the efficiency gain provided by the mixture Kalman filter can be very substantial. Another contribution of this article is the formulation of many nonlinear systems into conditional or partial conditional linear form, to which the mixture Kalman filter can be...

### Citations

1058 | Novel approach to nonlinear/non-Gaussian Bayesian state estimation - Gordon, Salmond, et al. - 1993 |

715 |
Tracking and Data Association
- Bar-Shalom
- 1988
(Show Context)
Citation Context ...s an important task to both civilian and military surveillance systems, particularly when a radar, sonar, or optical sensor is operated in the present of clutter or when innovations are non-Gaussian (=-=Bar-Shalom and Fortmann, 1988-=-). We show three examples of target tracking using the MKF: (a) targets in the presence of random interference (clutter); (b) targets with non-Gaussian innovations; and (c) targets with maneuvering. 5... |

714 |
Applied Optimal Estimation
- Gelb, editor
- 1974
(Show Context)
Citation Context ...tures of individual problems, some generalizations of the Kalman filtering methodology to nonlinear systems can be effective. For example, a few well-known extensions are the extended Kalman filters (=-=Gelb, 1974-=-), the Gaussian sum filters (Anderson and Moore, 1979), and the iterated extended Kalman filters (Jazwinski, 1970). Most of these methods are based on local linear approximations of the nonlinear syst... |

706 | Stochastic Processes and Filtering Theory - Jazwinski - 1970 |

657 | On sequential Monte Carlo sampling methods for Bayesian filtering - Doucet, Godsill, et al. - 2000 |

513 | Filtering via simulation: Auxiliary particle
- Pitt, Shephard
- 1999
(Show Context)
Citation Context ...n many recent modifications and improvements on the method (Berzuini, Best, Gilks, and Larizza 1997; Carpenter, Clifford, and Fearnhead 1997; Doucet 1998; Hurzeler and Kunsch 1995; Liu and Chen 1995; =-=Pitt and Shephard 1999-=-). A sequential importance sampling (SIS) framework is proposed in Liu and Chen (1998) to unify and generalize these related techniques. In the following context, we refer to all these methods applied... |

412 |
Simulation and the Monte Carlo Method
- Rubinstein
- 1981
(Show Context)
Citation Context ...the trajectory of an indicator variable (vector), the system is Gaussian and linear, for which the Kalman filter can be used. Thus, by using the marginalization technique for Monte Carlo computation (=-=Rubinstein, 1981-=-), we derive a Monte Carlo filter that focuses its full attention on the space of indicator variable. We call this filter a mixture Kalman filter (MKF). By doing so we can drastically reduce Monte Car... |

322 |
On Gibbs sampling for state space models
- Carter, Kohn
- 1994
(Show Context)
Citation Context ...tractive aspects of the CDLM and the MKF, but they are limited in scope. Several Markov chain Monte Carlo algorithms for this type of models have been proposed (See Carlin, Stoffer, and Polson, 1992; =-=Carter and Kohn, 1994-=-). In particular, Carter and Kohn (1994) present an efficient Gibbs sampler in which the indicatorst is the only latent variable to be imputed and the state variable x t is explicitly integrated out v... |

206 | On sequential simulation-based methods for bayesian filtering
- Doucet
- 1998
(Show Context)
Citation Context ...t areas that require dynamic modeling. There have also been many recent modifications and improvements on the method (Berzuini, Best, Gilks, and Larizza 1997; Carpenter, Clifford, and Fearnhead 1997; =-=Doucet 1998-=-; Hurzeler and Kunsch 1995; Liu and Chen 1995; Pitt and Shephard 1999). A sequential importance sampling (SIS) framework is proposed in Liu and Chen (1998) to unify and generalize these related techni... |

187 |
Covariance structure of the Gibbs sampler with applications to the comparisons of estimators and augmentation schemes
- Liu, Wong, et al.
- 1994
(Show Context)
Citation Context ... p(x t j y t ). Whereas the MKF tries to sample in the indicator space instead, which is equivalent to marginalizing out the x t . This operation has been shown to improve a Gibbs sampling algorithm (=-=Liu, Wong, and Kong, 1994-=-). Its advantage in a usual importance sampling scheme is shown in MacEachern et al. (1998). In the MKF setting, there is no clear theory available at this point. Our limited experience shows that the... |

183 | Sequential imputations and bayesian missing data problems - Kong, Lui, et al. - 1994 |

157 | Improved particle filter for nonlinear problems - Carpenter, Clifford, et al. - 1999 |

126 | A Monte Carlo approach to nonnormal and nonlinear state-space modeling - Carlin, Polson, et al. - 1992 |

86 |
Blind deconvolution via sequential imputations
- Liu, Chen
- 1995
(Show Context)
Citation Context ...There have also been many recent modifications and improvements on the method (Berzuini, Best, Gilks, and Larizza 1997; Carpenter, Clifford, and Fearnhead 1997; Doucet 1998; Hurzeler and Kunsch 1995; =-=Liu and Chen 1995-=-; Pitt and Shephard 1999). A sequential importance sampling (SIS) framework is proposed in Liu and Chen (1998) to unify and generalize these related techniques. In the following context, we refer to a... |

71 | Dynamic conditional independence models and Markov chain Monte Carlo methods - Berzuini, Best, et al. - 1997 |

67 | Sequential importance sampling for nonparametric Bayes models: the next generation, Canadian - MacEachern, Clyde, et al. - 1999 |

49 |
Monte Carlo approximation for general state space model
- Hurzeler, Kunsch
- 1998
(Show Context)
Citation Context ...require dynamic modeling. There have also been many recent modifications and improvements on the method (Berzuini, Best, Gilks, and Larizza 1997; Carpenter, Clifford, and Fearnhead 1997; Doucet 1998; =-=Hurzeler and Kunsch 1995-=-; Liu and Chen 1995; Pitt and Shephard 1999). A sequential importance sampling (SIS) framework is proposed in Liu and Chen (1998) to unify and generalize these related techniques. In the following con... |

48 | 1970)On state estimation in switching environments - Ackerson, Fu |

45 | A new approach to linear ltering and prediction problems - Kalman - 1960 |

42 | A stochastic simulation Bayesian approach to multitarget tracking - Avitzour - 1995 |

40 | Random sampling approach to state estimation in switching environments, Automatica 13 - Akashi, Kumamoto - 1977 |

39 |
1982)Detection and estimation for abruptly changing systems
- Tugnait
(Show Context)
Citation Context ...mption, and then used in a Gaussian approximation of the posterior of x t . Their approach can be easily generalized to update a segment ( t\Gammak ; : : : ;st ) of the indicator process recursively (=-=Tugnait, 1982). In deal-=-ing with the same CDLM, Akashi and Kumamoto (1977) introduce essentially a sequential importance sampling method for the indicator process in which an "optimal" sampling distribution is used... |

39 | Mixture models, Monte Carlo, Bayesian updating and dynamic models, Computing Science and Statistics 24: 325–333 - West - 1993 |

37 | A noniterative sampling/importance resampling alternative to the data augmentation algorithm for creating a few imputations when the fraction of missing information is modest: the SIR algorithm. Discussion of - Rubin - 1987 |

27 | Nonlinear Filters: Estimation and Applications - Tanizaki - 1996 |

11 | 1994)Partial non-Gaussian state space - Shephard |

10 | Optimal Filtering. Englewood Cli€s - Chen, Anderson, et al. - 1979 |

8 | Monte Carlo lter and smoother for non-Gaussian nonlinear state space models - Kitagawa - 1996 |

7 | An improved particle lter for non-linear problems - Carpenter, ord, et al. - 1999 |

3 |
Stochastic Processess and Filtering Theory
- Jazwinski
- 1970
(Show Context)
Citation Context ...can be effective. For example, a few well-known extensions are the extended Kalman filters (Gelb, 1974), the Gaussian sum filters (Anderson and Moore, 1979), and the iterated extended Kalman filters (=-=Jazwinski, 1970-=-). Most of these methods are based on local linear approximations of the nonlinear system. More recently, researchers began to pay attention to a new class of filtering methods based on the sequential... |

3 | 1978)On the detection of target trajectories in a multi-target environment - Smith, Winter |

3 | 1986)Applying the Monte Carlo method for optimum estimation in systems with random structure - Svetnik |

2 | 1994)Covariance structure of the Gibbs sampler with applications to the comparisons of estimators and augmentation schemes - Liu, Wong, et al. |

1 |
Optimal filtering, Prentice-Hall Avitzour, D
- Anderson, Moore
- 1979
(Show Context)
Citation Context ...ralizations of the Kalman filtering methodology to nonlinear systems can be effective. For example, a few well-known extensions are the extended Kalman filters (Gelb, 1974), the Gaussian sum filters (=-=Anderson and Moore, 1979-=-), and the iterated extended Kalman filters (Jazwinski, 1970). Most of these methods are based on local linear approximations of the nonlinear system. More recently, researchers began to pay attention... |

1 | 1998)Monte Carlo approximations for general state-space models - rzeler, M, et al. |

1 | 1998)Sequential Monte Carlo methods for dynamic systems - unknown authors |