Accession Number : AD0712408

Title :   ON BEHAVIOR STRATEGY SOLUTIONS IN FINITE EXTENDED DECISION PROCESSES,

Corporate Author : PURDUE UNIV LAFAYETTE IND SCHOOL OF ELECTRICAL ENGINEERING

Personal Author(s) : Utgoff,V. A. ; Kashyap,R. L.

Report Date : JUN 1970

Pagination or Media Count : 191

Abstract : The problem of determination of best behavior strategies for a decision maker in a two-person, zero-sum, finite extended decision process is considered. In this problem, best strategies are defined to be those strategies which maximize a player's expected return against all possible strategies that can be employed by the other player, and the decision process is thus considered as a two-person, zero-sum, finite extended game. In the report then, techniques are developed for finding best behavior strategies on arbitrary information collection schemes. These techniques make it possible to search in any given game for strategies which are minimally complex but acceptably effective. A doctor-patient medical game is searched using these techniques, and behavior strategies are found for the doctor which are more than 90% as effective but less than 10% as complex as the best strategy which results when the doctor gathers and retains all the information defined to be available to him. A pursuit and evasion game is also studied and best behavior strategies are determined for the pursuer and the evader when both players gather and retain all the information defined to be available to each of them. (Author)

Descriptors :   (*GAME THEORY, DECISION THEORY), INFORMATION THEORY, TOPOLOGY, MEDICAL PERSONNEL, SURGERY, DECISION MAKING, PROBABILITY, PROBABILITY DENSITY FUNCTIONS, LOGISTICS, DYNAMIC PROGRAMMING, SET THEORY, ALGORITHMS

Subject Categories : Operations Research

Distribution Statement : APPROVED FOR PUBLIC RELEASE