Simulation Approaches to General Probabilistic Inference on Belief Networks

By Ross D. Shachter, Mark Alan Peot. Conference on Uncertainty in Artificial Intelligence, 2013

393 citations recorded by the source corpus.

A number of algorithms have been developed to solve probabilistic inference problems on belief networks. These algorithms can be divided into two main groups: exact techniques which exploit the conditional independence revealed when the graph structure is relatively sparse, and probabilistic sampling techniques which exploit the “conductance” of an embedded Markov chain when the conditional probabilities have non-extreme values. In this paper, we investigate a family of “forward” Monte Carlo sampling techniques similar to Logic Sampling [Henrion, 1988] which appear to perform well even in some multiplyconnected networks with extreme conditional probabilities, and thus would be generally applicable. We consider several enhancements which reduce the posterior variance using this approach and propose a framework and criteria for choosing when to use those enhancements.

Subfields: Artificial Intelligence, Management Science and Operations Research

View Paper: https://www.semanticscholar.org/paper/6049c0a410322aa74f362c0b749e812b7b556a78

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

Citations in the corpus

Learning Stochastic Inverses (2013, neutral): “A number of adaptive importance sampling algorithms have been proposed for Bayesian networks, including Shachter and Peot (1989), Cheng and Druzdzel (2000), Yuan and Druzdzel (2012), Yu and Van Engelen (2012), Hernandez et al. (1998), Salmeron et al. (2000), and Ortiz and Kaelbling (2000).”
Learning Bayesian networks: approaches and issues (2011, background): “The first of these were by Fung and Chang (1990) and Shachter and Peot (1990) who discussed a strategy called likelihood weighting, that does not discard evidence.”
Learning Bayesian Network Model Structure from Data (2003, background): “Approximate ones (Henrion, 1988; Fung and Chang, 1989; Schachter and Peot, 1989) are sample-based, and generate an artificial database of samples during the process of estimation (the generated samples are discarded immediately and only the count of those than matched the query is kept).”
Probabilistic models with unknown objects (2006, methodology): “There is a simple and elegant way to apply importance sampling to a BN, called likelihood weighting [Fung and Chang, 1990; Shachter and Peot, 1990].”
The challenge of poker (2002, background): “(3) Modifying these calculations based on individual opponents is the subject of Section 6.”
Pattern Recognition and Machine Learning (2006, positive): “This method can be further extended using self-importance sampling (Shachter and Peot, 1990) in which the importance sampling distribution is continually updated to reflect the current estimated posterior distribution.”
Bayesian Artificial Intelligence (2004, methodology): “The logic sampling method was developed by Henrion [106], while likelihood weighting was developed in [87, 248].”
Probabilistic Diagnosis Using a Reformulation of the INTERNIST-1/QMR Knowledge Base (1991, methodology): “Using this technique, called import ance sampling, the algorithm instan tiates the diseases not by their prior probabilities P (dt), but rather by any sampling distribution P' (dtJ [37].”