Accession Number:

ADA110255

Title:

A Preliminary Report on a Multi-Level Learning Technique Using Production Systems.

Descriptive Note:

Interim technical rept. 1 Jul 80-30 Jun 81,

Corporate Author:

STATE UNIV OF NEW YORK AT BUFFALO AMHERST DEPT OF COMPUTER SCIENCE

Personal Author(s):

Report Date:

1981-11-01

Pagination or Media Count:

11.0

Abstract:

A generalized production system is described. The rules in it connect causally and stochastically related open and hidden variables. An open variable can be observed and measured at any time whereas the values of a hidden variable can be identified only at certain times. Both the causes and effects can be either open or hidden variables. The objective of the system is to provide sharper and sharper estimates of the values of hidden variables. A multi-level learning mechanism is outlined that leads to better and larger knowledge bases. Author

Subject Categories:

  • Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE