Fusion of Simultaneous Equation Models and Bayesian Networks

Prof. Marek Druzdzel, University of Pittsburgh
Tuesday, 15th April 2025

Simultaneous systems of equation have been in wide use in science and engineering for centuries. When based on structural equations, they can represent the causal structure of the underlying system, as explicated by Simon in 1953. This, in turn, is essential in intelligent systems. At this BNMA April Webinar, Professor Marek Druzdzel will describe his long-standing efforts to create a fusion between equation-based models and BNs. Marek will show how static BNs correspond to systems of structural equations and how dynamic BNs correspond to systems of simultaneous difference equations. Fusion of the two modelling formalisms allows for flexible representation of models consisting of discrete and continuous variables, common in domains such as engineering, economics, and risk analysis. Most of the ideas that he will present have been implemented in GeNIe, graphical modelling software developed originally in his research group, the Decision Systems Laboratory, University of Pittsburgh, and currently at BayesFusion, LLC (https://www.bayesfusion.com).

Marek Druzdzel is a Professor Emeritus in the School of Computing and Information, University of Pittsburgh. He holds a faculty appointment in the Computer Science Department, Bialystok University of Technology, Poland. His research interests focus on probabilistic and decision theoretic methods in decision support systems and human aspect of decision support (https://www.aragorn.wi.pb.edu.pl/~druzdzel).