Accession Number : AD0615119

Title :   BAYES ESTIMATION FOR SOME STIMULUS SAMPLING MODELS.

Descriptive Note : Technical memo.,

Corporate Author : SYSTEM DEVELOPMENT CORP SANTA MONICA CALIF

Personal Author(s) : Dear,Robert E.

Report Date : 22 FEB 1965

Pagination or Media Count : 46

Abstract : In this paper, the author derives the joint Bayes estimators of the three structural parameters of the single-element stimulus-sampling model of learning when conjugate prior beta distributions over these parameters are assumed. These estimators are obtained for finite sample sequences in contrast to the other available estimation procedures for this model which are based on infinite sample sequences. Extension of the Bayes estimation procedures to a two-element stimulussampling model involving six structural parameters is also carried out. The completeness of the class of Bayes estimators for these two models is established. Computational problems in obtaining values of the Bayes estimates are discussed. Although the estimators are functions which involve ratios of products and sums of many terms involving complete beta functions suitable approximate evaluations of the estimators can easily be carried out on a large computer. (Author)

Descriptors :   (*LEARNING, STATISTICAL ANALYSIS), (*CONDITIONED RESPONSE, LEARNING), MATHEMATICAL MODELS, DISTRIBUTION THEORY, SAMPLING, PROBABILITY

Distribution Statement : APPROVED FOR PUBLIC RELEASE