A hybrid connectionist expert system to solve classification problems

Describes HYCONES, a hybrid connectionist expert system model to solve classificational problems. The proposed model combines the formalism of frames with neural networks, taking advantage of this integration. While the symbolic paradigm provides rich and flexible constructs to describe the domain knowledge, the connectionist one endows the system with inductive and deductive learning capabilities. HYCONES architecture is presented, highlighting the hybrid knowledge base and its automatic knowledge acquisition technique from a case database. A prototype implementation to solve diagnostic problems in congenital heart diseases is presented, depicting its validation. The present status of the project and its future developments are discussed.<<ETX>>