AND/OR importance sampling
Pages 212 - 219
Abstract
The paper introduces AND/OR importance sampling for probabilistic graphical models. In contrast to importance sampling, AND/OR importance sampling caches samples in the AND/OR space and then extracts a new sample mean from the stored samples. We prove that AND/OR importance sampling may have lower variance than importance sampling; thereby providing a theoretical justification for preferring it over importance sampling. Our empirical evaluation demonstrates that AND/OR importance sampling is far more accurate than importance sampling in many cases.
References
[1]
[Bidyuk and Dechter, 2003] Bidyuk, B. and Dechter, R. (2003). An empirical study of w-cutset sampling for bayesian networks. In Proceedings of the 19th Annual Conference on Uncertainty in Artificial Intelligence (UAI-03).
[2]
[Cheng and Druzdzel, 2000] Cheng, J. and Druzdzel, M. J. (2000). Ais-bn: An adaptive importance sampling algorithm for evidential reasoning in large bayesian networks. J. Artif. Intell. Res. (JAIR), 13:155-188.
[3]
[Dawid et al., 1994] Dawid, A. P., Kjaerulff, U., and Lauritzen, S. L. (1994). Hybrid propagation in junction trees. In IPMU'94, pages 87-97.
[4]
[Dechter and Mateescu, 2007] Dechter, R. and Mateescu, R. (2007). AND/OR search spaces for graphical models. Artificial Intelligence, 171(2-3):73-106.
[5]
[Fishelson and Geiger, 2003] Fishelson, M. and Geiger, D. (2003). Optimizing exact genetic linkage computations. In RECOMB 2003.
[6]
[Geweke, 1989] Geweke, J. (1989). Bayesian inference in econometric models using monte carlo integration. Econometrica, 57(6):1317-39.
[7]
[Gogate and Dechter, 2005] Gogate, V. and Dechter, R. (2005). Approximate inference algorithms for hybrid bayesian networks with discrete constraints. UAI-2005.
[8]
[Gogate and Dechter, 2007] Gogate, V. and Dechter, R. (2007). Samplesearch: A scheme that searches for consistent samples. AISTATS 2007.
[9]
[Hernndez and Moral, 1995] Hernndez, L. D. and Moral, S. (1995). Mixing exact and importance sampling propagation algorithms in dependence graphs. International Journal of Approximate Reasoning, 12(8):553-576.
[10]
[Kjæerulff, 1995] Kjæerulff, U. (1995). Hugs: Combining exact inference and gibbs sampling in junction trees. In UAI, pages 368-375.
[11]
[Paskin, 2004] Paskin, M. A. (2004). Sample propagation. In Thrun, S., Saul, L., and Schölkopf, B., editors, Advances in Neural Information Processing Systems 16. MIT Press, Cambridge, MA.
[12]
[Rubinstein, 1981] Rubinstein, R. Y. (1981). Simulation and the Monte Carlo Method. John Wiley & Sons, Inc., New York, NY, USA.
[13]
[Yuan and Druzdzel, 2006] Yuan, C. and Druzdzel, M. J. (2006). Importance sampling algorithms for Bayesian networks: Principles and performance. Mathematical and Computer Modelling.
- AND/OR importance sampling
Recommendations
Importance sampling: a review
AbstractWe provide a short overview of importance sampling—a popular sampling tool used for Monte Carlo computing. We discuss its mathematical foundation and properties that determine its accuracy in Monte Carlo approximations. We review the fundamental ...
Table-driven adaptive importance sampling
EGSR '08: Proceedings of the Nineteenth Eurographics conference on RenderingMonte Carlo rendering algorithms generally rely on some form of importance sampling to evaluate the measurement equation. Most of these importance sampling methods only take local information into account, however, so the actual importance function used ...
Incremental wavelet importance sampling for direct illumination
VRST '07: Proceedings of the 2007 ACM symposium on Virtual reality software and technologyMost of existing importance sampling methods for direct illumination exploit importance of illumination and surface BRDF. Without taking the visibility into consideration, they can not adaptively adjust the number of samples for each pixel during the ...
Comments
Information & Contributors
Information
Published In
Publisher
AUAI Press
Arlington, Virginia, United States
Publication History
Published: 09 July 2008
Qualifiers
- Article
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 0Total Downloads
- Downloads (Last 12 months)0
- Downloads (Last 6 weeks)0
Reflects downloads up to 06 Jan 2025