In an apparatus and method for computerized design of fuzzy rules from training data vectors, a membership function type is selected from a set of predetermined membership function types, and membership values for the training data vectors and center of gravity values for the membership functions are determined in alternation in an iterative method for all training data vectors and for each membership function. When an abort criterion is met, then the most recently determined membership functions, described by the center of gravity values, are employed as rules.

 
Web www.patentalert.com

< Computer-aided methods and apparatus for assessing an organizational process or system

< Finding collective baskets and inference rules for internet or intranet mining for large data bases

> Method and system for reducing memory requirements by using separate specification and storage of default and overriding attributes

> Efficient fuzzy logic fault accommodation algorithm

~ 00206