A knowledge-based system for multiple pattern recognition paradigms support

This paper focuses on developing a knowledge-based system with multiple pattern recognition paradigms to support a variety of facilities for an intelligent pattern classifier design. Since re-developing a pattern recognition system to fulfill the new requirements can be quite expensive, our work has inspired with research in software re-engineering by treating the existing pattern recognition software as a valuable resource. The approach lies in extracting knowledge intended to be reused from original pattern recognition programs and replacing them with libraries containing pre-generated packages. We then integrate these transformed packages into a specific application.