A study of automatic rule generation and adaption

Publisher Website
Google Scholar
Abstract
This paper describes a rule-based learning system called CSM which acquires knowledge through trial-and-error interaction automatically and builds its knowledge base incrementally without human assistance. The development of CSM suggests a promising way to bridge the gap between general-purpose, low-level adaptive learning systems and specialized symbolic rule-based AI systems. To show the feasibility of this approach, CSM has been tested both in the robot navigation domain and letter extrapolation domain. The experimental results show that CSM is able to create appropriate rules from scratch to solve given tasks in an increasingly efficient and direct manner.

This publication has 0 references indexed in Scilit: