Strong & Weak Methods: A Logical View of Uncertainty

TL;DR

This paper discusses extending classical probability frameworks in AI to explicitly represent various types of uncertainty, including possibility and plausibility, and advocates for weak methods and symbolic uncertainty calculus.

Contribution

It introduces three arguments for broadening the probability framework without challenging classical methods, emphasizing explicit representation and flexible uncertainty management.

Findings

Explicit representation of possibility and plausibility.

Use of weak methods in poorly defined problems.

Symbolic representation of different uncertainty calculi.

Abstract

The last few years has seen a growing debate about techniques for managing uncertainty in AI systems. Unfortunately this debate has been cast as a rivalry between AI methods and classical probability based ones. Three arguments for extending the probability framework of uncertainty are presented, none of which imply a challenge to classical methods. These are (1) explicit representation of several types of uncertainty, specifically possibility and plausibility, as well as probability, (2) the use of weak methods for uncertainty management in problems which are poorly defined, and (3) symbolic representation of different uncertainty calculi and methods for choosing between them.