Soft Evidential Update for Probabilistic Multiagent Systems

Marco Valtorta, PhD
Department of Computer Science and Engineering
University of South Carolina
mgv@cse.sc.edu

Thursday, June 14, 2001
11AM-12PM,  126-225

Abstract

We address the problem of updating a probability distribution
represented by a Bayesian network upon presentation of soft evidence.
Our motivation is the desire to let agents communicate with each other
by exchanging beliefs represented by probability distributions, as in
the Agent Encapsulated Bayesian Network (AEBN) model. We explain how to
model soft evidence with observation variables in a way that insures
that the independence structure of the model is maintained even upon
receipt of soft evidence. We describe the big clique algorithm, which
extends the celebrated junction tree algorithm to soft evidential update
and has been implemented with Young-Gyun Kim in the BC-Hugin program.
To illustrate the use of BC-Hugin we present a hypothetical mission
scenario involving a diagnostic agent module that updates its belief in
the fault of a component upon receipt of soft evidence from a second
(possibly remote) diagnostic agent.

Biography

Dr. Marco Valtorta is an Associate Professor of Computer Science and
Engineering and Director of the Artificial Intelligence Lab at the
University of South Carolina. He received his PhD in Computer Science
from Duke University in 1987.  Between 1985 and 1988, he was a Project
Officer for ESPRIT at the Commission of the European Communities in
Brussels, where he supervised projects in the Advanced Information
Processing area.
Dr. Valtorta is a member of AAAI, ACM, IEEE, and Sigma Xi.  He is the
author of 75 technical publications. His research interests are in the
areas of normative reasoning under uncertainty (especially Bayesian
networks, influence diagrams, and their use in stand-alone and
multiagent systems), heuristics for problem solving, and computational
complexity in artificial intelligence.  He is an Associate Editor of the
International Journal of Approximate Reasoning and a member of the
editorial board of Applied Intelligence.  He was a Lilly teaching fellow
in 1993-94.