Machine learning models for expert systems
Abstract
Current expert systems have been successfully designed for several application domains to solve difficult problems that ordinarily require expertise. However, there are still some major problems with current expert systems. They cannot acquire or generate new knowledge inductively, and they cannot improve on their performance, either. In this research, a modified memory-based learning algorithm and a memory-based inductive learning algorithm have been devised to solve the knowledge acquisition problem, while planning and local strategies have been devised to equip current expert systems for self-skill refinement. Examples have been used to show our learning models are workable. Finally, contributions and limitations of this research have been discussed. Also, worthwhile further research has been pointed out and a conceptual framework from the socioeconomic point of view has been proposed.
Degree
Ph.D.
Advisors
Whinston, Purdue University.
Subject Area
Management
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.