A connectionist production system with approximate matching function

A hybrid symbolic/connectionist implementation of a production system with an approximate matching function is presented. The authors demonstrate the conversion of sequential production rules into a parallel firing connectionist network consisting of simple cells and providing an approximate matching function. The hybrid symbolic/connectionist production system (HSC-PS) supports two types of working memory element, the attribute/value pair and the object/attribute/value triplet. An attribute may be either crisp or fuzzy. The degree of match is measured for a condition having a fuzzy attribute and the working memory element which exists for that attribute and is represented by a value between 0 and 1. The strength of firing for a rule is decided by the minimum match of all the conditions in the left hand side of the rule. The strength of firing affects the actual amount of action to be performed on a fuzzy attribute. If two or more actions are to be performed on the same fuzzy attribute, a new action is composed from them and executed. A classic control problem is taken as an example and the results of the HSC-PS are compared with those of a fuzzy logic controller