Refractor importance sampling

Haohai Yu, Robert A. van Engelen
Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence, PMLR R6:603-609, 2008.

Abstract

In this paper we introduce Refractor Importance Sampling (RIS), an improvement to reduce error variance in Bayesian network importance sampling propagation under evidential reasoning. We prove the existence of a collection of importance functions that are close to the optimal importance function under evidential reasoning. Based on this theoretic result we derive the RIS algorithm. RIS approaches the optimal importance function by applying localized arc changes to minimize the divergence between the evidence-adjusted importance function and the optimal importance function. The validity and performance of RIS is empirically tested with a large set of synthetic Bayesian networks and two real-world networks.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR6-yu08a, title = {Refractor importance sampling}, author = {Yu, Haohai and van Engelen, Robert A.}, booktitle = {Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence}, pages = {603--609}, year = {2008}, editor = {McAllester, David A. and Myllymäki, Petri}, volume = {R6}, series = {Proceedings of Machine Learning Research}, month = {09--12 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r6/main/assets/yu08a/yu08a.pdf}, url = {https://proceedings.mlr.press/r6/yu08a.html}, abstract = {In this paper we introduce Refractor Importance Sampling (RIS), an improvement to reduce error variance in Bayesian network importance sampling propagation under evidential reasoning. We prove the existence of a collection of importance functions that are close to the optimal importance function under evidential reasoning. Based on this theoretic result we derive the RIS algorithm. RIS approaches the optimal importance function by applying localized arc changes to minimize the divergence between the evidence-adjusted importance function and the optimal importance function. The validity and performance of RIS is empirically tested with a large set of synthetic Bayesian networks and two real-world networks.}, note = {Reissued by PMLR on 09 October 2024.} }
Endnote
%0 Conference Paper %T Refractor importance sampling %A Haohai Yu %A Robert A. van Engelen %B Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2008 %E David A. McAllester %E Petri Myllymäki %F pmlr-vR6-yu08a %I PMLR %P 603--609 %U https://proceedings.mlr.press/r6/yu08a.html %V R6 %X In this paper we introduce Refractor Importance Sampling (RIS), an improvement to reduce error variance in Bayesian network importance sampling propagation under evidential reasoning. We prove the existence of a collection of importance functions that are close to the optimal importance function under evidential reasoning. Based on this theoretic result we derive the RIS algorithm. RIS approaches the optimal importance function by applying localized arc changes to minimize the divergence between the evidence-adjusted importance function and the optimal importance function. The validity and performance of RIS is empirically tested with a large set of synthetic Bayesian networks and two real-world networks. %Z Reissued by PMLR on 09 October 2024.
APA
Yu, H. & van Engelen, R.A.. (2008). Refractor importance sampling. Proceedings of the 24th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R6:603-609 Available from https://proceedings.mlr.press/r6/yu08a.html. Reissued by PMLR on 09 October 2024.

Related Material