Accession Number:

AD0625048

Title:

A GENERAL THEORY OF FINITE SEQUENTIAL DECISION PROCESSES.

Descriptive Note:

Technical paper,

Corporate Author:

RESEARCH ANALYSIS CORP MCLEAN VA

Personal Author(s):

Report Date:

1965-02-01

Pagination or Media Count:

32.0

Abstract:

This paper discusses the structure and analysis of sequential decision processes within the restriction of a formulation involving only finite sets. A definition of the phrase finite sequential decision process is presented, using the notions of action sets, outcome sets, conditional probabilities, histories, and availability functions. To the best of our knowledge this formulation is more general than any presented to date, and it is designed to fill an existing gap in the literature of sequential decision processes. The latter part of the paper examines special conditions of homogeneity, availabilities, probabilities, and utilities that reduce the general process to special cases, some of which have been discussed in the literature. Along the way a general expected-utility decision model is presented for the general finite process. The model is based on utilities of terminal histories and probabilities of terminal histories conditioned by strategies. The definition of strategy is based on the notion of decision functions for the various stages of the process. A backward-computational approach for obtaining optimal decision functions is discussed. Author

Subject Categories:

  • Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE