SYNTHESIS OF CASCADE RECURRENT NEURAL NETWORKS USING FEEDFORWARD GENERALIZATION PROPERTIES

This paper presents a new analysis and synthesis technique for a class of recurrent networks known as a Cascade Recurrent Network (CRN). In this technique, a feedforward (FF) sub-network is used with synchronou s feedback to implement associative memory (AM). FF network mapping pr operties are shown to determine CRN stability, and, on the basis of th is stability-mapping relation, a new synthesis technique is given. Thi s technique utilizes the optimization of the FF mapping sub-network ge neralization as a synthesis procedure for CRN. Sample results are show n. (C) 1998 Elsevier Science Inc, All rights reserved.