Sparse Bayesian Estimation of Parameters in Linear-Gaussian State-Space Models

被引:3
|
作者
Cox, Benjamin [1 ]
Elvira, Victor [1 ]
机构
[1] Univ Edinburgh, Sch Math, Edinburgh EH9 3FD, Scotland
基金
英国自然环境研究理事会;
关键词
Bayesian methods; graphical inference; Kalman filtering; Markov chain Monte Carlo; parameter estimation; sparsity detection; state-space modelling; REVERSIBLE-JUMP; WEAK-CONVERGENCE; PARTICLE FILTER; SELECTION; ALGORITHM; INFERENCE; DECOMPOSITION; COMPUTATION; LIKELIHOOD;
D O I
10.1109/TSP.2023.3278867
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
State-space models (SSMs) are a powerful statistical tool for modelling time-varying systems via a latent state. In these models, the latent state is never directly observed. Instead, a sequence of data points related to the state are obtained. The linear-Gaussian state-space model is widely used, since it allows for exact inference when all model parameters are known, however this is rarely the case. The estimation of these parameters is a very challenging but essential task to perform inference and prediction. In the linear-Gaussian model, the state dynamics are described via a state transition matrix. This model parameter is known to behard to estimate, since it encodes the relationships between the state elements, which are never observed. In many applications, this transition matrix is sparse since not all state components directly affect all other state components. However, most parameter estimation methods do not exploit this feature. In this work we propose SpaRJ, a fully probabilistic Bayesian approach that obtains sparse samples from the posterior distribution of the transition matrix. Our method explores sparsity by traversing a set of models that exhibit differing sparsity patterns in the transition matrix. Moreover, we also design new effective rules to explore transition matrices within the same level of sparsity. This novel methodology has strong theoretical guarantees, and unveils the latent structure of the data generating process, thereby enhancing interpretability. The performance of SpaRJ is showcased in example with dimension 144 in the parameter space, and in a numerical example with real data.
引用
收藏
页码:1922 / 1937
页数:16
相关论文
共 50 条
  • [31] Sequential Estimation of Gaussian Process-Based Deep State-Space Models
    Liu, Yuhao
    Ajirak, Marzieh
    Djuric, Petar M.
    IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2023, 71 : 2968 - 2980
  • [32] State-space estimation with uncertain models
    Sayed, AH
    Subramanian, A
    TOTAL LEAST SQUARES AND ERRORS-IN-VARIABLES MODELING: ANALYSIS, ALGORITHMS AND APPLICATIONS, 2002, : 191 - 202
  • [33] Distributed and Recursive Parameter Estimation in Parametrized Linear State-Space Models
    Ram, S. Sundhar
    Veeravalli, Venugopal V.
    Nedic, Angelia
    IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 2010, 55 (02) : 488 - 492
  • [34] A Sparse Bayesian Approach to the Identification of Nonlinear State-Space Systems
    Pan, Wei
    Yuan, Ye
    Goncalves, Jorge
    Stan, Guy-Bart
    IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 2016, 61 (01) : 182 - 187
  • [35] On the latent state estimation of nonlinear population dynamics using Bayesian and non-Bayesian state-space models
    Wang, Guiming
    ECOLOGICAL MODELLING, 2007, 200 (3-4) : 521 - 528
  • [36] Annealed MCMC for Bayesian Learning of Linear Gaussian State Space Models with Changepoints
    Bunch, Pete
    Murphy, James
    Godsill, Simon
    2016 19TH INTERNATIONAL CONFERENCE ON INFORMATION FUSION (FUSION), 2016, : 597 - 605
  • [37] Estimation of Granger Causality of State-Space Models using a Clustering with Gaussian Mixture Model
    Plub-in, Nattaporn
    Songsiri, Jitkomut
    2019 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN AND CYBERNETICS (SMC), 2019, : 3853 - 3858
  • [38] Online Bayesian Tire-Friction Learning by Gaussian-Process State-Space Models
    Berntorp, Karl
    IFAC PAPERSONLINE, 2020, 53 (02): : 13939 - 13944
  • [39] Laplace Approximated Gaussian Process State-Space Models
    Lindinger, Jakob
    Rakitsch, Barbara
    Lippert, Christoph
    UNCERTAINTY IN ARTIFICIAL INTELLIGENCE, VOL 180, 2022, 180 : 1199 - 1209
  • [40] BOOTSTRAPPING STATE-SPACE MODELS - GAUSSIAN MAXIMUM-LIKELIHOOD-ESTIMATION AND THE KALMAN FILTER
    STOFFER, DS
    WALL, KD
    JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1991, 86 (416) : 1024 - 1033