Belief networks have become an increasingly popular mechanism for dealing with uncertainty in systems. Unfortunately, it is known that finding the probability values of belief network nodes given a set of evidence is not tractable in general. Many different simulation algorithms for approximating solutions to this problem have been proposed and implemented. In this report, we describe the implementation of a collection of such algorithms, CABeN. CABeN contains a library of routines for simulating belief networks, a program for accessing the routines through menus on any 'tty' interface, and some sample programs demonstrating how the library would be used within an application. CABeN implements five algorithms: Logic Sampling, Likelihood, Weighting (Shachter's Basic algorithm), Self Importance, Pearl's algorithm, and Chavez's algorithm. In addition, we have implemented Markov scoring as an option to any of the above algorithms. We have compared these 10 variations with each other in a series of experiments in which we varied the graph topologies, the number of nodes provided with evidence, and the conditional probability values. A detailed description of each... Read complete abstract on page 2.
[1]
Mark E. Frisse,et al.
Query Networks for Medical Information Retrieval-Assigning Probabilistic Relationships
,
1990
.
[2]
David J. Spiegelhalter,et al.
Local computations with probabilities on graphical structures and their application to expert systems
,
1990
.
[3]
Ross D. Shachter,et al.
Simulation Approaches to General Probabilistic Inference on Belief Networks
,
2013,
UAI.
[4]
Gregory F. Cooper,et al.
Current research directions in the development of expert systems based on belief networks
,
1989
.
[5]
Max Henrion,et al.
An Introduction to Algorithms for Inference in Belief Nets
,
1989,
UAI.