โ Scribed by J.J. Buckley; W. Siler; Douglas Tucker
No coin nor oath required. For personal study only.
We describe a fuzzy rule based expert production system. The system accepts as input a fuzzy vector all of whose components are fuzzy sets, and produces as output a tizzy set of conclusions. Non-fuzzy data are stored as fuzzy data with grade of membership one; internally, all data are considered fuzzy. Data are processed by a set of rules, each containing a list of patterns to be matched against current data, with consequent actions to be taken when a sufficient match is found. A fuzzy set of tireable rules is defined for each state of the system's working memory; this set is executed recursively until a null alpha-cut set of fireable rules is encountered. An application is given of the system to a problem in medical image processing.
๐ SIMILAR VOLUMES
We argue that with the addition of a rule-ranking procedure, our fuzzy expert system FLOPS can correct previous conclusions given new, possibly conflicting, information.