Approximation of hidden Markov models by mixtures of experts with application to particle filtering

Jimmy Olsson, Jonas Ströjby
Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR 9:573-580, 2010.

Abstract

Selecting conveniently the proposal kernel and the adjustment multiplier weights of the auxiliary particle filter may increase significantly the accuracy and computational efficiency of the method. However, in practice the optimal proposal kernel and multiplier weights are seldom known. In this paper we present a simulation-based method for constructing offline an approximation of these quantities that makes the filter close to fully adapted at a reasonable computational cost. The approximation is constructed as a mixture of experts optimised through an efficient stochastic approximation algorithm. The method is illustrated on two simulated examples.

Cite this Paper

BibTeX
@InProceedings{pmlr-v9-olsson10a, title = {Approximation of hidden Markov models by mixtures of experts with application to particle filtering}, author = {Olsson, Jimmy and Ströjby, Jonas}, booktitle = {Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics}, pages = {573--580}, year = {2010}, editor = {Teh, Yee Whye and Titterington, Mike}, volume = {9}, series = {Proceedings of Machine Learning Research}, address = {Chia Laguna Resort, Sardinia, Italy}, month = {13--15 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v9/olsson10a/olsson10a.pdf}, url = {https://proceedings.mlr.press/v9/olsson10a.html}, abstract = {Selecting conveniently the proposal kernel and the adjustment multiplier weights of the auxiliary particle filter may increase significantly the accuracy and computational efficiency of the method. However, in practice the optimal proposal kernel and multiplier weights are seldom known. In this paper we present a simulation-based method for constructing offline an approximation of these quantities that makes the filter close to fully adapted at a reasonable computational cost. The approximation is constructed as a mixture of experts optimised through an efficient stochastic approximation algorithm. The method is illustrated on two simulated examples.} }
Endnote
%0 Conference Paper %T Approximation of hidden Markov models by mixtures of experts with application to particle filtering %A Jimmy Olsson %A Jonas Ströjby %B Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2010 %E Yee Whye Teh %E Mike Titterington %F pmlr-v9-olsson10a %I PMLR %P 573--580 %U https://proceedings.mlr.press/v9/olsson10a.html %V 9 %X Selecting conveniently the proposal kernel and the adjustment multiplier weights of the auxiliary particle filter may increase significantly the accuracy and computational efficiency of the method. However, in practice the optimal proposal kernel and multiplier weights are seldom known. In this paper we present a simulation-based method for constructing offline an approximation of these quantities that makes the filter close to fully adapted at a reasonable computational cost. The approximation is constructed as a mixture of experts optimised through an efficient stochastic approximation algorithm. The method is illustrated on two simulated examples.
RIS
TY - CPAPER TI - Approximation of hidden Markov models by mixtures of experts with application to particle filtering AU - Jimmy Olsson AU - Jonas Ströjby BT - Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics DA - 2010/03/31 ED - Yee Whye Teh ED - Mike Titterington ID - pmlr-v9-olsson10a PB - PMLR DP - Proceedings of Machine Learning Research VL - 9 SP - 573 EP - 580 L1 - http://proceedings.mlr.press/v9/olsson10a/olsson10a.pdf UR - https://proceedings.mlr.press/v9/olsson10a.html AB - Selecting conveniently the proposal kernel and the adjustment multiplier weights of the auxiliary particle filter may increase significantly the accuracy and computational efficiency of the method. However, in practice the optimal proposal kernel and multiplier weights are seldom known. In this paper we present a simulation-based method for constructing offline an approximation of these quantities that makes the filter close to fully adapted at a reasonable computational cost. The approximation is constructed as a mixture of experts optimised through an efficient stochastic approximation algorithm. The method is illustrated on two simulated examples. ER -
APA
Olsson, J. & Ströjby, J.. (2010). Approximation of hidden Markov models by mixtures of experts with application to particle filtering. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 9:573-580 Available from https://proceedings.mlr.press/v9/olsson10a.html.

Related Material