Controlling chaos in chaotic neural networks

✍ Scribed by Shin Mizutani; Takuya Sano; Tadasu Uchiyama; Noboru Sonehara

• Publisher: John Wiley and Sons
• Year: 1998
• Tongue: English
• Weight: 277 KB
• Volume: 81
• Category: Article
• ISSN: 1042-0967
• DOI: 10.1002/(sici)1520-6440(199808)81:8<73::aid-ecjc8>3.0.co;2-0

✦ Synopsis

This work demonstrates the control of chaos in chaotic neural networks. Chaotic neural networks, which were proposed by Aihara and others, consist of chaotic neuron models, and are based on research on the giant axon of squids and study of the Hodgkin-Huxley equation. They show a chaotic response that cannot be expressed by conventional neuron models. Although research has been performed on utilizing this chaotic response for active search of associative memory, it is difficult to determine when to stop the chaotic dynamics. Therefore, we have recently investigated chaotic control methods that had been the subject of previous research. Among these control methods, there is a method in which unstable period points are stabilized by multiplying the exponent or exponential function by the system parameters. We used an improved version of this method for high-dimension system control. For simplicity, we describe the control of networks that are coupled with the first order nearest neighbors and have global homogeneous coupling. We confirm that by using this control method, the unstable period points in the network can be stabilized and control of chaos is possible. With this method, stabilization is also possible when noise is added to the system.