Genetic classifier system approach in knowledge-based system
- Bangkok : Asian Institute of Technology, 1993
- 54 leaves + 1 online resource
- Thesis ; no. CS-93-30 .
- Asian Institute of Technology. Thesis ; no. CS-93-30 .
A thesis submitted in partial fulfillment of the requirement for the degree of Master of Science
Thesis (M.Sc.) - Asian Institute of Technology, 1993
To build up a knowledge-based system in Artificial Intelligence (Al), selecting an appropriate set of rules is one of the key problems. In this thesis, a Genetic Classifier System approach is employed to optimize the rules for knowledge-based system. The performance of a classifier system which belongs to the genetics-based machine learning architecture is tested using the various Genetic Algorithm operators, namely: reproduction, crossover, and mutation. Together with these GA operators, rule/message and apportionment of credit determine the fittest set of string rules, expressed as classifiers which serve as solution set, in two cases, namely: the eleven-multiplexer task and the Expert System of stock-cutting problem. Based on the experimental results, it is suggested that Genetic Classifier System is a feasible approach to optimize the rules for improving the performance of knowledge-based system.