An expert system for pattern recognition based on features and knowledge

Abstract

An expert system for pattern recognition based on features and knowledge is realized, and the theoretical and practical problems involved are discussed. The system, designed for analysis of geological patterns, has two novel characteristics. First, the system uses both features and knowledge for its analysis; there is some interaction between the two types of data. Secondly, the control strategy is based on an incomplete-reasoning and regression thinking model. The system is realized and proved to be effective.<>