[comp.ai.digest] Seminar - Evidential Reasoning: Overview and Implementation

lunt@CSL.SRI.COM.UUCP (08/13/87)

SRI COMPUTER SCIENCE LAB SEMINAR SERIES ANNOUNCES:



	   EVIDENTIAL REASONING: OVERVIEW AND IMPLEMENTATION


			       TOM GARVEY

	              AI CENTER, SRI INTERNATIONAL

                      Monday, August 17 at 4:00 pm
          SRI International, Computer Science Laboratory, BN182


Evidential reasoning consists of theoretical and practical methods for
reasoning from evidence, the uncertain, imprecise, and sometimes
incorrect information that is typically provided by "real-world"
information sources.  This theory evolved in response to the apparent
representational and computational inadequacies of classical probability
methods when dealing with such information in an expert system
framework.  Evidential reasoning is (currently) theoretically grounded
in the Shafer-Dempster theory of evidence.  Using this theory, we have
developed procedures for fusing multiple, distinct bodies of evidence,
for projecting evidential statements in time, for translating statements
in one vocabulary into a different one, for interpreting selected
propositions based on a given body evidence, and for summarizing and
"gisting" a body of evidence.  This seminar will be in the nature of a
high-level tutorial describing the Shafer-Dempster theory and Gister,
our current implementation of evidential reasoning.


-------