New Results on the Compatibility of the Standard Conditional Likelihood Approach with Product Space Conditional Event Algebra

Abstract

It is an understatement that information in general arrives in conditional form. In fact, unconditional information can always be considered as the special case where the antecedent is universal. Nevertheless, it is surprising how little progress has been made in developing a full logic of conditional expressions, compatible with traditional conditional probability techniques. Recently, this problem has begun to be addressed by a number of researchers, who have commonly agreed that, as a basis, conditional events (a/b) , where a and be are ordinary unconditional ones lying in some sample space (boolean or sigma-algebra), should be interpreted as intervals of events (or their logical equivalence) A*B, B' V A*B, relative to the usual event or set partial order less than or equal. Following this a number of conditional event algebras have been derived/proposed, possessing several desirable mathematical- logical properties and relatively easy to implement. (See, e.g. Goodman, Nguyen, and Walker, Conditional Inference and Logic for Intelligent Systems. A Theory of Measure-Free Conditioning, North-Holland, 1991.) However, a major drawback with these algebras is that inherently they cannot be boolean. (At most, they can have an algebraic structure which is a Stone algebra-see the above reference.) This leads, in particular, to certain problems in relating this work to the standard numerically-oriented approach of conditional probability, including the modeling of random variables and higher order conditional forms.... Data fusion, Semantics, Conditional events, Linguistic information, Syntactics.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 01, 1992
Accession Number
ADA262854

Entities

People

  • I. R. Goodman

Organizations

  • Naval Command, Control and Ocean Surveillance Center

Tags

DTIC Thesaurus Topics

  • Data Fusion
  • Intelligent Systems
  • Intervals
  • Military Research
  • Ocean Surveillance
  • Probability
  • Random Variables
  • Semantics
  • Standards
  • Surveillance

Fields of Study

  • Mathematics

Readers

  • Artificial Intelligence
  • Graph Algorithms and Convex Optimization.
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms
  • Space