Cellular neural network

DETAILED DESCRIPTION In accordance with the present invention, a new circuit architecture, called a cellular neural network, which can perform parallel signal processing in real time, is disclosed.
In view of the nearest neighbor interactive property of cellular neural networks, they are much more amenable to VLSI implementation than general neural networks.
In spite of the "local nature" of the nearest neighbor interconnections, cellular neural networks are nevertheless imbued with some global properties because of the propagation effects of the local interactions during the transient regime.
This transient regime can be exploited to show how the ability to analyze the "local" dynamics (via the dynamic route approach) will allow the system trajectories to be steered into a configuration of stable equilibria corresponding to some global pattern that is sought to be recognized.
Indeed, the ability to control the local circuit dynamics stands out as one of most desirable features of cellular neural networks.
Furthermore, cellular neural networks have practical dynamic ranges, whereas, general neural networks often suffer from severe dynamic range restrictions in the circuit implementation state.