A maximum entropy approach to nonmonotonic reasoning

Author

Goldszmidt, Moisés ; Morris, Paul ; Pearl, Judea

Author_Institution

Rockwell Int. Sci. Center, Palo Alto, CA, USA

Volume

15

Issue

3

fYear

1993

fDate

3/1/1993 12:00:00 AM

Firstpage

220

Lastpage

232

Abstract

An approach to nonmonotonic reasoning that combines the principle of infinitesimal probabilities with that of maximum entropy, thus extending the inferential power of the probabilistic interpretation of defaults, is proposed. A precise formalization of the consequences entailed by a conditional knowledge base is provided, the computational machinery necessary for drawing these consequences is developed, and the behavior of the maximum entropy approach is compared to related work in default reasoning. The resulting formalism offers a compromise between two extremes: the cautious approach based on the conditional interpretations of defaults and the bold approach based on minimizing abnormalities

Keywords

knowledge based systems; nonmonotonic reasoning; probabilistic logic; abnormality minimisation; conditional interpretations; conditional knowledge base; inference; infinitesimal probabilities; maximum entropy; nonmonotonic reasoning; probabilistic interpretation; Artificial intelligence; Birds; Computer science; Contracts; Entropy; Helium; Logic; Machinery; Minimization; NASA;

fLanguage

English

Journal_Title

Pattern Analysis and Machine Intelligence, IEEE Transactions on

Publisher

ieee

ISSN

0162-8828

Type

jour

DOI

10.1109/34.204904

Filename

204904