The design of CMOS nonself-feedback ratio memory cellular nonlinear network without elapsed operation for pattern learning and recognition

In this paper, a nonself-feedback ratio memory cellular nonlinear network (RMCNN) without elapsed operation is proposed and implemented in CMOS for image pattern learning and recognition. In the non-self-feedback RMCNN, the self-feedback template coefficient is not used and operation of non-self-feedback RMCNN can be simplified. The final weights of template A are generated directly after learning by incorporating the ratio-memory function into the learning rule so that no elapsed period is required for ratio memory. In the proposed RMCNN, simple multiplication circuit and comparator circuit are used and the chip area can be reduced. The pattern learning and recognition behavior of the proposed RMCNN is simulated by Matlab and Hspice. It is found that the performance is the same as that of RMCNN with elapsed operation.