Irregular-Time Bayesian Networks
Michael Ramati, Yuval Shahar
In many fields observations are performed ir- regularly along time, due to either measure- ment limitations or lack of a constant im- manent rate. While discrete-time Markov models (as Dynamic Bayesian Networks) in- troduce either inefficient computation or an information loss to reasoning about such processes, continuous-time Markov models assume either a discrete state space (as Continuous-Time Bayesian Networks), or a
at continuous state space (as stochastic dif- ferential equations). To address these prob- lems, we present a new modeling class called Irregular-Time Bayesian Networks (ITBNs), generalizing Dynamic Bayesian Networks, al- lowing substantially more compact represen- tations, and increasing the expressivity of the temporal dynamics. In addition, a globally optimal solution is guaranteed when learn- ing temporal systems, provided that they are fully observed at the same irregularly spaced time-points, and a semiparametric subclass of ITBNs is introduced to allow further adap- tation to the irregular nature of the available data.
PDF Link: /papers/10/p484-ramati.pdf
AUTHOR = "Michael Ramati
and Yuval Shahar",
TITLE = "Irregular-Time Bayesian Networks",
BOOKTITLE = "Proceedings of the Twenty-Sixth Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-10)",
PUBLISHER = "AUAI Press",
ADDRESS = "Corvallis, Oregon",
YEAR = "2010",
PAGES = "484--491"