Software for data analysis with graphical models

Wray L. Buntine RIACS at NASA Ames Research Center Mail Stop 269{2, Mo ett Field, CA 94035{1000, USA [email protected] H. Scott Roy Heuristicrats Research, Inc. 1678 Shattuck Avenue, Suite 310, Berkeley, CA 94709-1631, USA [email protected] Abstract Probabilistic graphical models are being used widely in arti cial intelligence and statistics, for instance, in diagnosis and expert systems, as a framework for representing and reasoning with probabilities and independencies. They come with corresponding algorithms for performing statistical inference. This o ers a unifying framework for prototyping and/or generating data analysis algorithms from graphical speci cations. This paper illustrates the framework with an example and then presents some basic techniques for the task: problem decomposition and the calculation of exact Bayes factors. Other tools already developed, such as automatic di erentiation, Gibbs sampling, and use of the EM algorithm, make this a broad basis for the generation of data analysis software.