Volume 19, Number 4, Winter 2011
A DISCRETE TIME DELAYED NEURAL
NETWORK WITH POTENTIAL FOR
ASSOCIATIVE MEMORY REVISITED
N. KOSOVALIĆ AND J. WU
Abstract. We consider a nonlinear difference equation with a time delay which represents the simplest possible discretely updated neural network. We show that provided that the integral time delay is large enough, this self feedback system exhibits a huge number of asymptotically stable periodic orbits. Therefore, such a network has potential for associative memory and pattern recognition.