Download Evidential Reasoning Under Uncertainty

Excerpt from "Exploring Artificial Intellioence·
Survey Talks from the National Conferences.on
Artificial Intelligence!' Ed. H. Shrobe,
Morgan-Kaufmann, 1988.
Chapter
10
Evidential Reasoning Under
Uncertainty
Judea Pearl
Cognitive Systems Laboratory
Computer Science Department
University of California, Los Angeles
1 Introduction
1.1 Overview
One can hardly identify a field in AI that doesn't use some sort of evidential
reasoning. namely, processes leading from evidence or clues to guesses and
conclusions under conditions of partial information. Therefore, to avoid having
to cover the entire field of AI, the topic will be limited to evidential reasoning
tasks in which the uncertainty is given a specific notation, namely, it is repre­
sented explicitly by some sort of measure or degree.
Constrained by this guideline, I will not be able to give a full account of
the heuristic approaches to evidential reasoning [Cohen, 1985; Clancey, 1985)
nor to works in truth-maintenance systems and nonmonotonic reasoning that,
essentially, address the same sort of problems. The latter are given full cover­
age by other surveys (sec this volume), and will only be touched on briefly to
point out their fundamental ties to other formalisms.
381
TECHNICAL REPORT
R-107-S