Evolving Fuzzy Logic Control Strategies using SAMUEL: An Initial Implementation.
Abstract
Many control systems have been successfully implemented using fuzzy logic, which provides a systematic method for reasoning about uncertainty using expressions found in natural language. This paper describes an extension of the SAMUEL learning system to include fuzzy logic. SAMUEL is a learning system that uses genetic algorithms and other learning methods to evolve refined rules from an initial set of rules provided by the user. In this initial implementation, SAMUEL searches for the rules making up a fuzzy knowledge base (that is, a control strategy), given the user's definition of the fuzzy variables, the values that the variables can take on, and the fixed membership functions associated with the fuzzy values. The genetic algorithm searches for the combinations of rules that make up effective strategies, including the level of generality expressed by the rules. An example is provided showing how to learn fuzzy rules for evasive maneuvers.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 06, 1996
- Accession Number
- ADA317388
Entities
People
- Helen G. Cobb
- John J. Grefenstette
Organizations
- United States Naval Research Laboratory