Bayesian joint state and parameter tracking in autoregressive models

Ismail Senoz, Albert Podusenko, Wouter M. Kouw, Bert Vries
Proceedings of the 2nd Conference on Learning for Dynamics and Control, PMLR 120:95-104, 2020.

Abstract

We address the problem of online Bayesian state and parameter tracking in autoregressive (AR) models with time-varying process noise variance. The involved marginalization and expectation integrals cannot be analytically solved. Moreover, the online tracking constraint makes sampling and batch learning methods unsuitable for this problem. We propose a hybrid variational message passing algorithm that robustly tracks the time-varying dynamics of the latent states, AR coefficients and process noise variance. Since message passing in a factor graph is a highly modular inference approach, the proposed methods easily extend to other non-stationary dynamic modeling problems.

Cite this Paper

BibTeX
@InProceedings{pmlr-v120-senoz20a, title = {Bayesian joint state and parameter tracking in autoregressive models}, author = {Senoz, Ismail and Podusenko, Albert and Kouw, Wouter M. and de Vries, Bert}, booktitle = {Proceedings of the 2nd Conference on Learning for Dynamics and Control}, pages = {95--104}, year = {2020}, editor = {Bayen, Alexandre M. and Jadbabaie, Ali and Pappas, George and Parrilo, Pablo A. and Recht, Benjamin and Tomlin, Claire and Zeilinger, Melanie}, volume = {120}, series = {Proceedings of Machine Learning Research}, month = {10--11 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v120/senoz20a/senoz20a.pdf}, url = {https://proceedings.mlr.press/v120/senoz20a.html}, abstract = {We address the problem of online Bayesian state and parameter tracking in autoregressive (AR) models with time-varying process noise variance. The involved marginalization and expectation integrals cannot be analytically solved. Moreover, the online tracking constraint makes sampling and batch learning methods unsuitable for this problem. We propose a hybrid variational message passing algorithm that robustly tracks the time-varying dynamics of the latent states, AR coefficients and process noise variance. Since message passing in a factor graph is a highly modular inference approach, the proposed methods easily extend to other non-stationary dynamic modeling problems.} }
Endnote
%0 Conference Paper %T Bayesian joint state and parameter tracking in autoregressive models %A Ismail Senoz %A Albert Podusenko %A Wouter M. Kouw %A Bert Vries %B Proceedings of the 2nd Conference on Learning for Dynamics and Control %C Proceedings of Machine Learning Research %D 2020 %E Alexandre M. Bayen %E Ali Jadbabaie %E George Pappas %E Pablo A. Parrilo %E Benjamin Recht %E Claire Tomlin %E Melanie Zeilinger %F pmlr-v120-senoz20a %I PMLR %P 95--104 %U https://proceedings.mlr.press/v120/senoz20a.html %V 120 %X We address the problem of online Bayesian state and parameter tracking in autoregressive (AR) models with time-varying process noise variance. The involved marginalization and expectation integrals cannot be analytically solved. Moreover, the online tracking constraint makes sampling and batch learning methods unsuitable for this problem. We propose a hybrid variational message passing algorithm that robustly tracks the time-varying dynamics of the latent states, AR coefficients and process noise variance. Since message passing in a factor graph is a highly modular inference approach, the proposed methods easily extend to other non-stationary dynamic modeling problems.
APA
Senoz, I., Podusenko, A., Kouw, W.M. & Vries, B.. (2020). Bayesian joint state and parameter tracking in autoregressive models. Proceedings of the 2nd Conference on Learning for Dynamics and Control, in Proceedings of Machine Learning Research 120:95-104 Available from https://proceedings.mlr.press/v120/senoz20a.html.

Related Material