City Research Online

Abstract

This thesis investigates the addition of experience learning components to types of general problem solver which have been advocated by the Artificial Intelligence community for use in planning domains. The learning components considered preserve the general applicability of a problem solver while allowing for it to improve it’s efficiency when applied to a particular domain. Various heuristic acquisition methods are presented, as well as three types of problem solver; together they have all been implemented in a large integrated system called "FM".

A specific aim is to demonstrate that a particular planning and learning configuration can significantly improve its efficiency by the automatic acquisition of strong heuristics, using a novel heuristic aquisition method. The body of the thesis concentrates on this particular configuration which proved successful in a range of planning applications.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Departments:	School of Science & Technology > Department of Computer Science School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads