Author

Fuller

Date of Award

4-1984

Degree Name

Master of Science

Department

Computer Science

Access Setting

Masters Thesis-Open Access

Abstract

A learning system is examined which is capable of learning a finite state machine from a class of finite state machines based on the observed behavior of the machine. The size of the search space becomes very large as the number of states in the machine increases. The size of the search space quickly becomes the limiting factor in the size of the class of machines which may be learned. An investigation is made of methods to improve the performance of the learning system. The application of a depth first approach to the development of the search space is shown to provide a significant reduction in the amount of space required to identify a machine. An heuristic estimator is used to guide the depth first search and is shown to have a potential for reducing the time required to identify the machine.

Share

COinS