A novel GCM chaotic neural network for information processing

A new chaotic neural network named “globally coupled map using sine map(SI-GCM)”, which is a modified Kaneko’s globally coupled map model, is proposed. With the introduction of sine map and chaotic neurons’ different way of coupling, it exhibits rich dynamic behaviors. By adopting a variable threshold parameter control method, it can be controlled to specified-period orbit. Furthermore, the controlled SI-GCM has excellent associative memory performance. It can not only output unique fixed pattern, but also output periodic patterns which contain the stored pattern closest to the initial pattern. Simulation results suggest that SI-GCM is fit for information processing.

► A new GCM chaotic neural network model named SI-GCM is proposed. ► The way of delay coupling is presented for SI-GCM. ► The SI-GCM can be controlled to expected periodic orbits. ► The dynamical associative memory of the controlled SI-GCM is illustrated. ► The controlled SI-GCM can output whether unique fixed pattern or periodic patterns.